Day Day Paper-Reading
Computation and Language 98
☆ Understanding the Limits of Lifelong Knowledge Editing in LLMs
Keeping large language models factually up-to-date is crucial for deployment, yet costly retraining remains a challenge. Knowledge editing offers a promising alternative, but methods are only tested on small-scale or synthetic edit benchmarks. In this work, we aim to bridge research into lifelong knowledge editing to real-world edits at practically relevant scale. We first introduce WikiBigEdit; a large-scale benchmark of real-world Wikidata edits, built to automatically extend lifelong for future-proof benchmarking. In its first instance, it includes over 500K question-answer pairs for knowledge editing alongside a comprehensive evaluation pipeline. Finally, we use WikiBigEdit to study existing knowledge editing techniques' ability to incorporate large volumes of real-world facts and contrast their capabilities to generic modification techniques such as retrieval augmentation and continual finetuning to acquire a complete picture of the practical extent of current lifelong knowledge editing.
comment: Preprint
☆ Symbolic Mixture-of-Experts: Adaptive Skill-based Routing for Heterogeneous Reasoning
Combining existing pre-trained expert LLMs is a promising avenue for scalably tackling large-scale and diverse tasks. However, selecting experts at the task level is often too coarse-grained, as heterogeneous tasks may require different expertise for each instance. To enable adaptive instance-level mixing of pre-trained LLM experts, we propose Symbolic-MoE, a symbolic, text-based, and gradient-free Mixture-of-Experts framework. Symbolic-MoE takes a fine-grained approach to selection by emphasizing skills, e.g., algebra in math or molecular biology in biomedical reasoning. We propose a skill-based recruiting strategy that dynamically selects the most relevant set of expert LLMs for diverse reasoning tasks based on their strengths. Each selected expert then generates its own reasoning, resulting in k outputs from k experts, which are then synthesized into a final high-quality response by an aggregator chosen based on its ability to integrate diverse reasoning outputs. We show that Symbolic-MoE's instance-level expert selection improves performance by a large margin but -- when implemented naively -- can introduce a high computational overhead due to the need for constant model loading and offloading. To address this, we implement a batch inference strategy that groups instances based on their assigned experts, loading each model only once. This allows us to integrate 16 expert models on 1 GPU with a time cost comparable to or better than prior multi-agent baselines using 4 GPUs. Through extensive evaluations on diverse benchmarks (MMLU-Pro, GPQA, AIME, and MedMCQA), we demonstrate that Symbolic-MoE outperforms strong LLMs like GPT4o-mini, as well as multi-agent approaches, with an absolute average improvement of 8.15% over the best multi-agent baseline. Moreover, Symbolic-MoE removes the need for expensive multi-round discussions, outperforming discussion baselines with less computation.
comment: The first three authors contributed equally. Project Page: https://symbolic_moe.github.io/
☆ Learning LLM Preference over Intra-Dialogue Pairs: A Framework for Utterance-level Understandings
Large language models (LLMs) have demonstrated remarkable capabilities in handling complex dialogue tasks without requiring use case-specific fine-tuning. However, analyzing live dialogues in real-time necessitates low-latency processing systems, making it impractical to deploy models with billions of parameters due to latency constraints. As a result, practitioners often prefer smaller models with millions of parameters, trained on high-quality, human-annotated datasets. Yet, curating such datasets is both time-consuming and costly. Consequently, there is a growing need to combine the scalability of LLM-generated labels with the precision of human annotations, enabling fine-tuned smaller models to achieve both higher speed and accuracy comparable to larger models. In this paper, we introduce a simple yet effective framework to address this challenge. Our approach is specifically designed for per-utterance classification problems, which encompass tasks such as intent detection, dialogue state tracking, and more. To mitigate the impact of labeling errors from LLMs -- the primary source of inaccuracies in student models -- we propose a noise-reduced preference learning loss. Experimental results demonstrate that our method significantly improves accuracy across utterance-level dialogue tasks, including sentiment detection (over $2\%$), dialogue act classification (over $1.5\%$), etc.
comment: 7 pages, 4 figures
☆ A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models
Large Language Models (LLMs) have revolutionized natural language processing, yet their internal mechanisms remain largely opaque. Recently, mechanistic interpretability has attracted significant attention from the research community as a means to understand the inner workings of LLMs. Among various mechanistic interpretability approaches, Sparse Autoencoders (SAEs) have emerged as a particularly promising method due to their ability to disentangle the complex, superimposed features within LLMs into more interpretable components. This paper presents a comprehensive examination of SAEs as a promising approach to interpreting and understanding LLMs. We provide a systematic overview of SAE principles, architectures, and applications specifically tailored for LLM analysis, covering theoretical foundations, implementation strategies, and recent developments in sparsity mechanisms. We also explore how SAEs can be leveraged to explain the internal workings of LLMs, steer model behaviors in desired directions, and develop more transparent training methodologies for future models. Despite the challenges that remain around SAE implementation and scaling, they continue to provide valuable tools for understanding the internal mechanisms of large language models.
comment: 20 pages, 3 figures
☆ R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning
Existing Large Reasoning Models (LRMs) have shown the potential of reinforcement learning (RL) to enhance the complex reasoning capabilities of Large Language Models~(LLMs). While they achieve remarkable performance on challenging tasks such as mathematics and coding, they often rely on their internal knowledge to solve problems, which can be inadequate for time-sensitive or knowledge-intensive questions, leading to inaccuracies and hallucinations. To address this, we propose \textbf{R1-Searcher}, a novel two-stage outcome-based RL approach designed to enhance the search capabilities of LLMs. This method allows LLMs to autonomously invoke external search systems to access additional knowledge during the reasoning process. Our framework relies exclusively on RL, without requiring process rewards or distillation for a cold start. % effectively generalizing to out-of-domain datasets and supporting both Base and Instruct models. Our experiments demonstrate that our method significantly outperforms previous strong RAG methods, even when compared to the closed-source GPT-4o-mini.
☆ Quantifying the Robustness of Retrieval-Augmented Language Models Against Spurious Features in Grounding Data
Robustness has become a critical attribute for the deployment of RAG systems in real-world applications. Existing research focuses on robustness to explicit noise (e.g., document semantics) but overlooks spurious features (a.k.a. implicit noise). While previous works have explored spurious features in LLMs, they are limited to specific features (e.g., formats) and narrow scenarios (e.g., ICL). In this work, we statistically confirm the presence of spurious features in the RAG paradigm, a robustness problem caused by the sensitivity of LLMs to semantic-agnostic features. Moreover, we provide a comprehensive taxonomy of spurious features and empirically quantify their impact through controlled experiments. Further analysis reveals that not all spurious features are harmful and they can even be beneficial sometimes. Extensive evaluation results across multiple LLMs suggest that spurious features are a widespread and challenging problem in the field of RAG. The code and dataset will be released to facilitate future research. We release all codes and data at: $\\\href{https://github.com/maybenotime/RAG-SpuriousFeatures}{https://github.com/maybenotime/RAG-SpuriousFeatures}$.
☆ Evaluating open-source Large Language Models for automated fact-checking
The increasing prevalence of online misinformation has heightened the demand for automated fact-checking solutions. Large Language Models (LLMs) have emerged as potential tools for assisting in this task, but their effectiveness remains uncertain. This study evaluates the fact-checking capabilities of various open-source LLMs, focusing on their ability to assess claims with different levels of contextual information. We conduct three key experiments: (1) evaluating whether LLMs can identify the semantic relationship between a claim and a fact-checking article, (2) assessing models' accuracy in verifying claims when given a related fact-checking article, and (3) testing LLMs' fact-checking abilities when leveraging data from external knowledge sources such as Google and Wikipedia. Our results indicate that LLMs perform well in identifying claim-article connections and verifying fact-checked stories but struggle with confirming factual news, where they are outperformed by traditional fine-tuned models such as RoBERTa. Additionally, the introduction of external knowledge does not significantly enhance LLMs' performance, calling for more tailored approaches. Our findings highlight both the potential and limitations of LLMs in automated fact-checking, emphasizing the need for further refinements before they can reliably replace human fact-checkers.
comment: Main: 10 pages, 13 figures. Supplementary Materials: 7 pages, 29 figures, 1 table ### This work has been submitted to the IEEE for possible publication. ###
☆ Pi-GPS: Enhancing Geometry Problem Solving by Unleashing the Power of Diagrammatic Information
Geometry problem solving has garnered increasing attention due to its potential applications in intelligent education field. Inspired by the observation that text often introduces ambiguities that diagrams can clarify, this paper presents Pi-GPS, a novel framework that unleashes the power of diagrammatic information to resolve textual ambiguities, an aspect largely overlooked in prior research. Specifically, we design a micro module comprising a rectifier and verifier: the rectifier employs MLLMs to disambiguate text based on the diagrammatic context, while the verifier ensures the rectified output adherence to geometric rules, mitigating model hallucinations. Additionally, we explore the impact of LLMs in theorem predictor based on the disambiguated formal language. Empirical results demonstrate that Pi-GPS surpasses state-of-the-art models, achieving a nearly 10\% improvement on Geometry3K over prior neural-symbolic approaches. We hope this work highlights the significance of resolving textual ambiguity in multimodal mathematical reasoning, a crucial factor limiting performance.
☆ Cognitive Bias Detection Using Advanced Prompt Engineering
Cognitive biases, systematic deviations from rationality in judgment, pose significant challenges in generating objective content. This paper introduces a novel approach for real-time cognitive bias detection in user-generated text using large language models (LLMs) and advanced prompt engineering techniques. The proposed system analyzes textual data to identify common cognitive biases such as confirmation bias, circular reasoning, and hidden assumption. By designing tailored prompts, the system effectively leverages LLMs' capabilities to both recognize and mitigate these biases, improving the quality of human-generated content (e.g., news, media, reports). Experimental results demonstrate the high accuracy of our approach in identifying cognitive biases, offering a valuable tool for enhancing content objectivity and reducing the risks of biased decision-making.
comment: 17 pages. 6 Figures, 2 Tables
☆ Statistical Guarantees of Correctness Coverage for Medical Multiple-Choice Question Answering
Large language models (LLMs) are increasingly deployed in real-world question-answering (QA) applications. However, LLMs have been proven to generate hallucinations and nonfactual information, undermining their trustworthiness in high-stakes medical tasks. Conformal prediction (CP) is well-known to be model-agnostic and distribution-free, which creates statistically rigorous prediction sets in classification tasks. In this work, we for the first time adapt the CP framework to medical multiple-choice question-answering (MCQA) tasks, by correlating the nonconformity score with the frequency score of correct options grounded in self-consistency theory, assuming no access to internal model information. Considering that the adapted CP framework can only control the (mis)coverage rate, we employ a risk control framework, which can manage task-specific metrics by devising a monotonically decreasing loss function. We evaluate our framework on 3 popular medical MCQA datasets utilizing 4 ``off-the-shelf'' LLMs. Empirical results demonstrate that we achieve user-specified average (or marginal) error rates on the test set. Furthermore, we observe that the average prediction set size (APSS) on the test set decreases as the risk level increases, which concludes a promising evaluation metric for the uncertainty of LLMs.
comment: Under Review
☆ EuroBERT: Scaling Multilingual Encoders for European Languages
General-purpose multilingual vector representations, used in retrieval, regression and classification, are traditionally obtained from bidirectional encoder models. Despite their wide applicability, encoders have been recently overshadowed by advances in generative decoder-only models. However, many innovations driving this progress are not inherently tied to decoders. In this paper, we revisit the development of multilingual encoders through the lens of these advances, and introduce EuroBERT, a family of multilingual encoders covering European and widely spoken global languages. Our models outperform existing alternatives across a diverse range of tasks, spanning multilingual capabilities, mathematics, and coding, and natively supporting sequences of up to 8,192 tokens. We also examine the design decisions behind EuroBERT, offering insights into our dataset composition and training pipeline. We publicly release the EuroBERT models, including intermediate training checkpoints, together with our training framework.
comment: 26 pages, 6 figures, 11 tables
☆ Benchmarking LLMs in Recommendation Tasks: A Comparative Evaluation with Conventional Recommenders
In recent years, integrating large language models (LLMs) into recommender systems has created new opportunities for improving recommendation quality. However, a comprehensive benchmark is needed to thoroughly evaluate and compare the recommendation capabilities of LLMs with traditional recommender systems. In this paper, we introduce RecBench, which systematically investigates various item representation forms (including unique identifier, text, semantic embedding, and semantic identifier) and evaluates two primary recommendation tasks, i.e., click-through rate prediction (CTR) and sequential recommendation (SeqRec). Our extensive experiments cover up to 17 large models and are conducted across five diverse datasets from fashion, news, video, books, and music domains. Our findings indicate that LLM-based recommenders outperform conventional recommenders, achieving up to a 5% AUC improvement in the CTR scenario and up to a 170% NDCG@10 improvement in the SeqRec scenario. However, these substantial performance gains come at the expense of significantly reduced inference efficiency, rendering the LLM-as-RS paradigm impractical for real-time recommendation environments. We aim for our findings to inspire future research, including recommendation-specific model acceleration methods. We will release our code, data, configurations, and platform to enable other researchers to reproduce and build upon our experimental results.
☆ KIEval: Evaluation Metric for Document Key Information Extraction
Document Key Information Extraction (KIE) is a technology that transforms valuable information in document images into structured data, and it has become an essential function in industrial settings. However, current evaluation metrics of this technology do not accurately reflect the critical attributes of its industrial applications. In this paper, we present KIEval, a novel application-centric evaluation metric for Document KIE models. Unlike prior metrics, KIEval assesses Document KIE models not just on the extraction of individual information (entity) but also of the structured information (grouping). Evaluation of structured information provides assessment of Document KIE models that are more reflective of extracting grouped information from documents in industrial settings. Designed with industrial application in mind, we believe that KIEval can become a standard evaluation metric for developing or applying Document KIE models in practice. The code will be publicly available.
☆ Linear-MoE: Linear Sequence Modeling Meets Mixture-of-Experts
Linear Sequence Modeling (LSM) like linear attention, state space models and linear RNNs, and Mixture-of-Experts (MoE) have recently emerged as significant architectural improvements. In this paper, we introduce Linear-MoE, a production-level system for modeling and training large-scale models that integrate LSM with MoE. Linear-MoE leverages the advantages of both LSM modules for linear-complexity sequence modeling and MoE layers for sparsely activation, aiming to offer high performance with efficient training. The Linear-MoE system comprises: 1) Modeling subsystem, which provides a unified framework supporting all instances of LSM. and 2) Training subsystem, which facilitates efficient training by incorporating various advanced parallelism technologies, particularly Sequence Parallelism designed for Linear-MoE models. Additionally, we explore hybrid models that combine Linear-MoE layers with standard Transformer-MoE layers with its Sequence Parallelism to further enhance model flexibility and performance. Evaluations on two model series, A0.3B-2B and A1B-7B, demonstrate Linear-MoE achieves efficiency gains while maintaining competitive performance on various benchmarks, showcasing its potential as a next-generation foundational model architecture. Code: https://github.com/OpenSparseLLMs/Linear-MoE.
comment: Technical report, 17 pages
☆ An Empirical Study of Conformal Prediction in LLM with ASP Scaffolds for Robust Reasoning
In this paper, we examine the use of Conformal Language Modelling (CLM) alongside Answer Set Programming (ASP) to enhance the performance of standard open-weight LLMs on complex multi-step reasoning tasks. Using the StepGame dataset, which requires spatial reasoning, we apply CLM to generate sets of ASP programs from an LLM, providing statistical guarantees on the correctness of the outputs. Experimental results show that CLM significantly outperforms baseline models that use standard sampling methods, achieving substantial accuracy improvements across different levels of reasoning complexity. Additionally, the LLM-as-Judge metric enhances CLM's performance, especially in assessing structurally and logically correct ASP outputs. However, calibrating CLM with diverse calibration sets did not improve generalizability for tasks requiring much longer reasoning steps, indicating limitations in handling more complex tasks.
☆ Multi Agent based Medical Assistant for Edge Devices
Large Action Models (LAMs) have revolutionized intelligent automation, but their application in healthcare faces challenges due to privacy concerns, latency, and dependency on internet access. This report introduces an ondevice, multi-agent healthcare assistant that overcomes these limitations. The system utilizes smaller, task-specific agents to optimize resources, ensure scalability and high performance. Our proposed system acts as a one-stop solution for health care needs with features like appointment booking, health monitoring, medication reminders, and daily health reporting. Powered by the Qwen Code Instruct 2.5 7B model, the Planner and Caller Agents achieve an average RougeL score of 85.5 for planning and 96.5 for calling for our tasks while being lightweight for on-device deployment. This innovative approach combines the benefits of ondevice systems with multi-agent architectures, paving the way for user-centric healthcare solutions.
☆ Leveraging Semantic Type Dependencies for Clinical Named Entity Recognition
Previous work on clinical relation extraction from free-text sentences leveraged information about semantic types from clinical knowledge bases as a part of entity representations. In this paper, we exploit additional evidence by also making use of domain-specific semantic type dependencies. We encode the relation between a span of tokens matching a Unified Medical Language System (UMLS) concept and other tokens in the sentence. We implement our method and compare against different named entity recognition (NER) architectures (i.e., BiLSTM-CRF and BiLSTM-GCN-CRF) using different pre-trained clinical embeddings (i.e., BERT, BioBERT, UMLSBert). Our experimental results on clinical datasets show that in some cases NER effectiveness can be significantly improved by making use of domain-specific semantic type dependencies. Our work is also the first study generating a matrix encoding to make use of more than three dependencies in one pass for the NER task.
☆ Shifting Perspectives: Steering Vector Ensembles for Robust Bias Mitigation in LLMs ACL 2025
We present a novel approach to bias mitigation in large language models (LLMs) by applying steering vectors to modify model activations in forward passes. We employ Bayesian optimization to systematically identify effective contrastive pair datasets across nine bias axes. When optimized on the BBQ dataset, our individually tuned steering vectors achieve average improvements of 12.2%, 4.7%, and 3.2% over the baseline for Mistral, Llama, and Qwen, respectively. Building on these promising results, we introduce Steering Vector Ensembles (SVE), a method that averages multiple individually optimized steering vectors, each targeting a specific bias axis such as age, race, or gender. By leveraging their collective strength, SVE outperforms individual steering vectors in both bias reduction and maintaining model performance. The work presents the first systematic investigation of steering vectors for bias mitigation, and we demonstrate that SVE is a powerful and computationally efficient strategy for reducing bias in LLMs, with broader implications for enhancing AI safety.
comment: Submitted to ACL 2025
☆ Chain of Strategy Optimization Makes Large Language Models Better Emotional Supporter
The growing emotional stress in modern society has increased the demand for Emotional Support Conversations (ESC). While Large Language Models (LLMs) show promise for ESC, they face two key challenges: (1) low strategy selection accuracy, and (2) preference bias, limiting their adaptability to emotional needs of users. Existing supervised fine-tuning (SFT) struggles to address these issues, as it rigidly trains models on single gold-standard responses without modeling nuanced strategy trade-offs. To overcome these limitations, we propose Chain-of-Strategy Optimization (CSO), a novel approach that optimizes strategy selection preferences at each dialogue turn. We first leverage Monte Carlo Tree Search to construct ESC-Pro, a high-quality preference dataset with turn-level strategy-response pairs. Training on ESC-Pro with CSO improves both strategy accuracy and bias mitigation, enabling LLMs to generate more empathetic and contextually appropriate responses. Experiments on LLaMA-3.1-8B, Gemma-2-9B, and Qwen2.5-7B demonstrate that CSO outperforms standard SFT, highlighting the efficacy of fine-grained, turn-level preference modeling in ESC.
comment: 19 pages, 9 figures, 15 tables
☆ Improving Hate Speech Classification with Cross-Taxonomy Dataset Integration ACL
Algorithmic hate speech detection faces significant challenges due to the diverse definitions and datasets used in research and practice. Social media platforms, legal frameworks, and institutions each apply distinct yet overlapping definitions, complicating classification efforts. This study addresses these challenges by demonstrating that existing datasets and taxonomies can be integrated into a unified model, enhancing prediction performance and reducing reliance on multiple specialized classifiers. The work introduces a universal taxonomy and a hate speech classifier capable of detecting a wide range of definitions within a single framework. Our approach is validated by combining two widely used but differently annotated datasets, showing improved classification performance on an independent test set. This work highlights the potential of dataset and taxonomy integration in advancing hate speech detection, increasing efficiency, and ensuring broader applicability across contexts.
comment: Accepted for publication at LaTeCH-CLfL 2025. The 9th Joint ACL Special Interest Group on Language Technologies for the Socio-Economic Sciences and Humanities (SIGHUM) Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature
☆ GEMA-Score: Granular Explainable Multi-Agent Score for Radiology Report Evaluation
Automatic medical report generation supports clinical diagnosis, reduces the workload of radiologists, and holds the promise of improving diagnosis consistency. However, existing evaluation metrics primarily assess the accuracy of key medical information coverage in generated reports compared to human-written reports, while overlooking crucial details such as the location and certainty of reported abnormalities. These limitations hinder the comprehensive assessment of the reliability of generated reports and pose risks in their selection for clinical use. Therefore, we propose a Granular Explainable Multi-Agent Score (GEMA-Score) in this paper, which conducts both objective quantification and subjective evaluation through a large language model-based multi-agent workflow. Our GEMA-Score parses structured reports and employs NER-F1 calculations through interactive exchanges of information among agents to assess disease diagnosis, location, severity, and uncertainty. Additionally, an LLM-based scoring agent evaluates completeness, readability, and clinical terminology while providing explanatory feedback. Extensive experiments validate that GEMA-Score achieves the highest correlation with human expert evaluations on a public dataset, demonstrating its effectiveness in clinical scoring (Kendall coefficient = 0.70 for Rexval dataset and Kendall coefficient = 0.54 for RadEvalX dataset). The anonymous project demo is available at: https://github.com/Zhenxuan-Zhang/GEMA_score.
☆ AutoIOT: LLM-Driven Automated Natural Language Programming for AIoT Applications
The advent of Large Language Models (LLMs) has profoundly transformed our lives, revolutionizing interactions with AI and lowering the barrier to AI usage. While LLMs are primarily designed for natural language interaction, the extensive embedded knowledge empowers them to comprehend digital sensor data. This capability enables LLMs to engage with the physical world through IoT sensors and actuators, performing a myriad of AIoT tasks. Consequently, this evolution triggers a paradigm shift in conventional AIoT application development, democratizing its accessibility to all by facilitating the design and development of AIoT applications via natural language. However, some limitations need to be addressed to unlock the full potential of LLMs in AIoT application development. First, existing solutions often require transferring raw sensor data to LLM servers, which raises privacy concerns, incurs high query fees, and is limited by token size. Moreover, the reasoning processes of LLMs are opaque to users, making it difficult to verify the robustness and correctness of inference results. This paper introduces AutoIOT, an LLM-based automated program generator for AIoT applications. AutoIOT enables users to specify their requirements using natural language (input) and automatically synthesizes interpretable programs with documentation (output). AutoIOT automates the iterative optimization to enhance the quality of generated code with minimum user involvement. AutoIOT not only makes the execution of AIoT tasks more explainable but also mitigates privacy concerns and reduces token costs with local execution of synthesized programs. Extensive experiments and user studies demonstrate AutoIOT's remarkable capability in program synthesis for various AIoT tasks. The synthesized programs can match and even outperform some representative baselines.
☆ Speculative Decoding for Multi-Sample Inference
We propose a novel speculative decoding method tailored for multi-sample reasoning scenarios, such as self-consistency and Best-of-N sampling. Our method exploits the intrinsic consensus of parallel generation paths to synthesize high-quality draft tokens without requiring auxiliary models or external databases. By dynamically analyzing structural patterns across parallel reasoning paths through a probabilistic aggregation mechanism, it identifies consensus token sequences that align with the decoding distribution. Evaluations on mathematical reasoning benchmarks demonstrate a substantial improvement in draft acceptance rates over baselines, while reducing the latency in draft token construction. This work establishes a paradigm shift for efficient multi-sample inference, enabling seamless integration of speculative decoding with sampling-based reasoning techniques.
☆ Dynamic Knowledge Integration for Evidence-Driven Counter-Argument Generation with Large Language Models
This paper investigates the role of dynamic external knowledge integration in improving counter-argument generation using Large Language Models (LLMs). While LLMs have shown promise in argumentative tasks, their tendency to generate lengthy, potentially unfactual responses highlights the need for more controlled and evidence-based approaches. We introduce a new manually curated dataset of argument and counter-argument pairs specifically designed to balance argumentative complexity with evaluative feasibility. We also propose a new LLM-as-a-Judge evaluation methodology that shows a stronger correlation with human judgments compared to traditional reference-based metrics. Our experimental results demonstrate that integrating dynamic external knowledge from the web significantly improves the quality of generated counter-arguments, particularly in terms of relatedness, persuasiveness, and factuality. The findings suggest that combining LLMs with real-time external knowledge retrieval offers a promising direction for developing more effective and reliable counter-argumentation systems.
☆ Fine-Grained Evaluation for Implicit Discourse Relation Recognition
Implicit discourse relation recognition is a challenging task in discourse analysis due to the absence of explicit discourse connectives between spans of text. Recent pre-trained language models have achieved great success on this task. However, there is no fine-grained analysis of the performance of these pre-trained language models for this task. Therefore, the difficulty and possible directions of this task is unclear. In this paper, we deeply analyze the model prediction, attempting to find out the difficulty for the pre-trained language models and the possible directions of this task. In addition to having an in-depth analysis for this task by using pre-trained language models, we semi-manually annotate data to add relatively high-quality data for the relations with few annotated examples in PDTB 3.0. The annotated data significantly help improve implicit discourse relation recognition for level-2 senses.
☆ Uncertainty-Aware Decoding with Minimum Bayes Risk ICLR 2025
Despite their outstanding performance in the majority of scenarios, contemporary language models still occasionally generate undesirable outputs, for example, hallucinated text. While such behaviors have previously been linked to uncertainty, there is a notable lack of methods that actively consider uncertainty during text generation. In this work, we show how Minimum Bayes Risk (MBR) decoding, which selects model generations according to an expected risk, can be generalized into a principled uncertainty-aware decoding method. In short, we account for model uncertainty during decoding by incorporating a posterior over model parameters into MBR's computation of expected risk. We show that this modified expected risk is useful for both choosing outputs and deciding when to abstain from generation and can provide improvements without incurring overhead. We benchmark different methods for learning posteriors and show that performance improves with prediction diversity. We release our code publicly.
comment: ICLR 2025 (Poster)
☆ Coreference as an indicator of context scope in multimodal narrative
We demonstrate that large multimodal language models differ substantially from humans in the distribution of coreferential expressions in a visual storytelling task. We introduce a number of metrics to quantify the characteristics of coreferential patterns in both human- and machine-written texts. Humans distribute coreferential expressions in a way that maintains consistency across texts and images, interleaving references to different entities in a highly varied way. Machines are less able to track mixed references, despite achieving perceived improvements in generation quality.
comment: 20 pages, 4 tables
☆ Similarity-Based Domain Adaptation with LLMs
Unsupervised domain adaptation leverages abundant labeled data from various source domains to generalize onto unlabeled target data. Prior research has primarily focused on learning domain-invariant features across the source and target domains. However, these methods often require training a model using source domain data, which is time-consuming and can limit model usage for applications with different source data. This paper introduces a simple framework that utilizes the impressive generalization capabilities of Large Language Models (LLMs) for target data annotation without the need of source model training, followed by a novel similarity-based knowledge distillation loss. Our extensive experiments on cross-domain text classification reveal that our framework achieves impressive performance, specifically, 2.44\% accuracy improvement when compared to the SOTA method.
☆ Revealing Hidden Mechanisms of Cross-Country Content Moderation with Natural Language Processing
The ability of Natural Language Processing (NLP) methods to categorize text into multiple classes has motivated their use in online content moderation tasks, such as hate speech and fake news detection. However, there is limited understanding of how or why these methods make such decisions, or why certain content is moderated in the first place. To investigate the hidden mechanisms behind content moderation, we explore multiple directions: 1) training classifiers to reverse-engineer content moderation decisions across countries; 2) explaining content moderation decisions by analyzing Shapley values and LLM-guided explanations. Our primary focus is on content moderation decisions made across countries, using pre-existing corpora sampled from the Twitter Stream Grab. Our experiments reveal interesting patterns in censored posts, both across countries and over time. Through human evaluations of LLM-generated explanations across three LLMs, we assess the effectiveness of using LLMs in content moderation. Finally, we discuss potential future directions, as well as the limitations and ethical considerations of this work. Our code and data are available at https://github.com/causalNLP/censorship
☆ ZOGRASCOPE: A New Benchmark for Property Graphs
Natural language interfaces to knowledge graphs have become increasingly important in recent years, enabling easy and efficient access to structured data. In particular property graphs have seen growing adoption. However, these kind of graphs remain relatively underrepresented in research, which has focused in large part on RDF-style graphs. As a matter of fact there is a lack of resources for evaluating systems on property graphs, with many existing datasets featuring relatively simple queries. To address this gap, we introduce ZOGRASCOPE, a benchmark designed specifically for the cypher query language. The benchmark includes a diverse set of manually annotated queries of varying complexity. We complement this paper with a set of experiments that test the performance of out-of-the-box LLMs of different sizes. Our experiments show that semantic parsing over graphs is still a challenging open problem that can not be solved by prompting LLMs alone.
☆ PhiloBERTA: A Transformer-Based Cross-Lingual Analysis of Greek and Latin Lexicons
We present PhiloBERTA, a cross-lingual transformer model that measures semantic relationships between ancient Greek and Latin lexicons. Through analysis of selected term pairs from classical texts, we use contextual embeddings and angular similarity metrics to identify precise semantic alignments. Our results show that etymologically related pairs demonstrate significantly higher similarity scores, particularly for abstract philosophical concepts such as epist\=em\=e (scientia) and dikaiosyn\=e (iustitia). Statistical analysis reveals consistent patterns in these relationships (p = 0.012), with etymologically related pairs showing remarkably stable semantic preservation compared to control pairs. These findings establish a quantitative framework for examining how philosophical concepts moved between Greek and Latin traditions, offering new methods for classical philological research.
☆ WritingBench: A Comprehensive Benchmark for Generative Writing
Recent advancements in large language models (LLMs) have significantly enhanced text generation capabilities, yet evaluating their performance in generative writing remains a challenge. Existing benchmarks primarily focus on generic text generation or limited in writing tasks, failing to capture the diverse requirements of high-quality written contents across various domains. To bridge this gap, we present WritingBench, a comprehensive benchmark designed to evaluate LLMs across 6 core writing domains and 100 subdomains, encompassing creative, persuasive, informative, and technical writing. We further propose a query-dependent evaluation framework that empowers LLMs to dynamically generate instance-specific assessment criteria. This framework is complemented by a fine-tuned critic model for criteria-aware scoring, enabling evaluations in style, format and length. The framework's validity is further demonstrated by its data curation capability, which enables 7B-parameter models to approach state-of-the-art (SOTA) performance. We open-source the benchmark, along with evaluation tools and modular framework components, to advance the development of LLMs in writing.
☆ MM-StoryAgent: Immersive Narrated Storybook Video Generation with a Multi-Agent Paradigm across Text, Image and Audio
The rapid advancement of large language models (LLMs) and artificial intelligence-generated content (AIGC) has accelerated AI-native applications, such as AI-based storybooks that automate engaging story production for children. However, challenges remain in improving story attractiveness, enriching storytelling expressiveness, and developing open-source evaluation benchmarks and frameworks. Therefore, we propose and opensource MM-StoryAgent, which creates immersive narrated video storybooks with refined plots, role-consistent images, and multi-channel audio. MM-StoryAgent designs a multi-agent framework that employs LLMs and diverse expert tools (generative models and APIs) across several modalities to produce expressive storytelling videos. The framework enhances story attractiveness through a multi-stage writing pipeline. In addition, it improves the immersive storytelling experience by integrating sound effects with visual, music and narrative assets. MM-StoryAgent offers a flexible, open-source platform for further development, where generative modules can be substituted. Both objective and subjective evaluation regarding textual story quality and alignment between modalities validate the effectiveness of our proposed MM-StoryAgent system. The demo and source code are available.
☆ Personalized Text Generation with Contrastive Activation Steering
Personalized text generation aims to infer users' writing style preferences from their historical texts and generate outputs that faithfully reflect these stylistic characteristics. Existing solutions primarily adopt two paradigms: retrieval-augmented generation (RAG) and parameter-efficient fine-tuning (PEFT). While these approaches have advanced the field, they suffer from two critical limitations: (1) the entanglement of content semantics and stylistic patterns in historical texts impedes accurate modeling of user-specific writing preferences; and (2) scalability challenges arising from both RAG's inference latency by retrieval operations and PEFT's parameter storage requirements for per user model. To overcome these limitations, we propose StyleVector, a training-free framework that disentangles and represents personalized writing style as a vector in LLM's activation space, enabling style-steered generation during inference without requiring costly retrieval or parameter storage. Comprehensive experiments demonstrate that our framework achieves a significant 8% relative improvement in personalized generation while reducing storage requirements by 1700 times over PEFT method.
☆ Knowledge Updating? No More Model Editing! Just Selective Contextual Reasoning
As real-world knowledge evolves, the information embedded within large language models (LLMs) can become outdated, inadequate, or erroneous. Model editing has emerged as a prominent approach for updating LLMs' knowledge with minimal computational costs and parameter changes. This approach typically identifies and adjusts specific model parameters associated with newly acquired knowledge. However, existing methods often underestimate the adverse effects that parameter modifications can have on broadly distributed knowledge. More critically, post-edit LLMs frequently struggle with multi-hop reasoning and continuous knowledge updates. Although various studies have discussed these shortcomings, there is a lack of comprehensive evaluation. In this paper, we provide an evaluation of ten model editing methods along four dimensions: reliability, generalization, locality, and portability. Results confirm that all ten popular model editing methods show significant shortcomings across multiple dimensions, suggesting model editing is less promising. We then propose a straightforward method called Selective Contextual Reasoning (SCR), for knowledge updating. SCR does not modify model parameters but harnesses LLM's inherent contextual reasoning capabilities utilizing the updated knowledge pieces. Under SCR, an LLM first assesses whether an incoming query falls within the scope of an external knowledge base. If it does, the relevant external knowledge texts are contextualized to enhance reasoning; otherwise, the query is answered directly. We evaluate SCR against the ten model editing methods on two counterfactual datasets with three backbone LLMs. Empirical results confirm the effectiveness and efficiency of contextual reasoning for knowledge updating.
☆ Path Pooling: Train-Free Structure Enhancement for Efficient Knowledge Graph Retrieval-Augmented Generation
Although Large Language Models achieve strong success in many tasks, they still suffer from hallucinations and knowledge deficiencies in real-world applications. Many knowledge graph-based retrieval-augmented generation (KG-RAG) methods enhance the quality and credibility of LLMs by leveraging structure and semantic information in KGs as external knowledge bases. However, these methods struggle to effectively incorporate structure information, either incurring high computational costs or underutilizing available knowledge. Inspired by smoothing operations in graph representation learning, we propose path pooling, a simple, train-free strategy that introduces structure information through a novel path-centric pooling operation. It seamlessly integrates into existing KG-RAG methods in a plug-and-play manner, enabling richer structure information utilization. Extensive experiments demonstrate that incorporating the path pooling into the state-of-the-art KG-RAG method consistently improves performance across various settings while introducing negligible additional cost. Code is coming soon at https://github.com/hrwang00/path-pooling.
☆ ORANSight-2.0: Foundational LLMs for O-RAN
Despite the transformative impact of Large Language Models (LLMs) across critical domains such as healthcare, customer service, and business marketing, their integration into Open Radio Access Networks (O-RAN) remains limited. This gap is primarily due to the absence of domain-specific foundational models, with existing solutions often relying on general-purpose LLMs that fail to address the unique challenges and technical intricacies of O-RAN. To bridge this gap, we introduce ORANSight-2.0 (O-RAN Insights), a pioneering initiative aimed at developing specialized foundational LLMs tailored for O-RAN. Built on 18 LLMs spanning five open-source LLM frameworks, ORANSight-2.0 fine-tunes models ranging from 1 to 70B parameters, significantly reducing reliance on proprietary, closed-source models while enhancing performance for O-RAN. At the core of ORANSight-2.0 is RANSTRUCT, a novel Retrieval-Augmented Generation (RAG) based instruction-tuning framework that employs two LLM agents to create high-quality instruction-tuning datasets. The generated dataset is then used to fine-tune the 18 pre-trained open-source LLMs via QLoRA. To evaluate ORANSight-2.0, we introduce srsRANBench, a novel benchmark designed for code generation and codebase understanding in the context of srsRAN, a widely used 5G O-RAN stack. We also leverage ORANBench13K, an existing benchmark for assessing O-RAN-specific knowledge. Our comprehensive evaluations demonstrate that ORANSight-2.0 models outperform general-purpose and closed-source models, such as ChatGPT-4o and Gemini, by 5.421% on ORANBench and 18.465% on srsRANBench, achieving superior performance while maintaining lower computational and energy costs. We also experiment with RAG-augmented variants of ORANSight-2.0 LLMs and thoroughly evaluate their energy characteristics, demonstrating costs for training, standard inference, and RAG-augmented inference.
☆ Memory-augmented Query Reconstruction for LLM-based Knowledge Graph Reasoning
Large language models (LLMs) have achieved remarkable performance on knowledge graph question answering (KGQA) tasks by planning and interacting with knowledge graphs. However, existing methods often confuse tool utilization with knowledge reasoning, harming readability of model outputs and giving rise to hallucinatory tool invocations, which hinder the advancement of KGQA. To address this issue, we propose Memory-augmented Query Reconstruction for LLM-based Knowledge Graph Reasoning (MemQ) to decouple LLM from tool invocation tasks using LLM-built query memory. By establishing a memory module with explicit descriptions of query statements, the proposed MemQ facilitates the KGQA process with natural language reasoning and memory-augmented query reconstruction. Meanwhile, we design an effective and readable reasoning to enhance the LLM's reasoning capability in KGQA. Experimental results that MemQ achieves state-of-the-art performance on widely used benchmarks WebQSP and CWQ.
☆ Rewarding Curse: Analyze and Mitigate Reward Modeling Issues for LLM Reasoning
Chain-of-thought (CoT) prompting demonstrates varying performance under different reasoning tasks. Previous work attempts to evaluate it but falls short in providing an in-depth analysis of patterns that influence the CoT. In this paper, we study the CoT performance from the perspective of effectiveness and faithfulness. For the former, we identify key factors that influence CoT effectiveness on performance improvement, including problem difficulty, information gain, and information flow. For the latter, we interpret the unfaithful CoT issue by conducting a joint analysis of the information interaction among the question, CoT, and answer. The result demonstrates that, when the LLM predicts answers, it can recall correct information missing in the CoT from the question, leading to the problem. Finally, we propose a novel algorithm to mitigate this issue, in which we recall extra information from the question to enhance the CoT generation and evaluate CoTs based on their information gain. Extensive experiments demonstrate that our approach enhances both the faithfulness and effectiveness of CoT.
comment: 18 pages, 21 figures
☆ Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching
Recent advances in large language models have demonstrated remarkable reasoning capabilities through Chain of Thought (CoT) prompting, but often at the cost of excessive verbosity in their intermediate outputs, which increases computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting framework that combines cognitive-inspired reasoning paradigms with linguistic constraints to minimize token usage while preserving reasoning accuracy. SoT is designed as a flexible framework that can incorporate any custom reasoning paradigms based on cognitive science, and we instantiate it with three such paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each tailored to different reasoning tasks and selected dynamically via a lightweight routing model. Through comprehensive evaluation across 15 reasoning datasets with multiple languages and multimodal scenarios, we demonstrate that SoT achieves token reductions of 76% with negligible accuracy impact. In certain domains like mathematical and multi-hop reasoning, it even improves accuracy while using significantly fewer tokens. Our code is publicly available: https://www.github.com/SimonAytes/SoT.
☆ Ensemble Debiasing Across Class and Sample Levels for Fairer Prompting Accuracy
Language models are strong few-shot learners and achieve good overall accuracy in text classification tasks, masking the fact that their results suffer from great class accuracy imbalance. We believe that the pursuit of overall accuracy should not come from enriching the strong classes, but from raising up the weak ones. To address the imbalance, we propose a post-hoc nonlinear integer programming based debiasing method that ensembles weight correction and membership correction to enable flexible rectifications of class probabilities at both class and sample levels, enhancing the performance of LLMs directly from their outputs. Evaluations with Llama-2-13B on seven text classification benchmarks show that our approach achieves state-of-the-art overall accuracy gains with balanced class accuracies. The resulted probability correction scheme demonstrates that sample-level corrections are necessary to elevate weak classes. In addition, due to effectively correcting weak classes, our method also brings significant performance gains to Llama-2-70B, especially on a biomedical domain task, demonstrating its effectiveness across both small and large model variants.
☆ Interpersonal Memory Matters: A New Task for Proactive Dialogue Utilizing Conversational History
Proactive dialogue systems aim to empower chatbots with the capability of leading conversations towards specific targets, thereby enhancing user engagement and service autonomy. Existing systems typically target pre-defined keywords or entities, neglecting user attributes and preferences implicit in dialogue history, hindering the development of long-term user intimacy. To address these challenges, we take a radical step towards building a more human-like conversational agent by integrating proactive dialogue systems with long-term memory into a unified framework. Specifically, we define a novel task named Memory-aware Proactive Dialogue (MapDia). By decomposing the task, we then propose an automatic data construction method and create the first Chinese Memory-aware Proactive Dataset (ChMapData). Furthermore, we introduce a joint framework based on Retrieval Augmented Generation (RAG), featuring three modules: Topic Summarization, Topic Retrieval, and Proactive Topic-shifting Detection and Generation, designed to steer dialogues towards relevant historical topics at the right time. The effectiveness of our dataset and models is validated through both automatic and human evaluations. We release the open-source framework and dataset at https://github.com/FrontierLabs/MapDia.
☆ RocketEval: Efficient Automated LLM Evaluation via Grading Checklist ICLR 2025
Evaluating large language models (LLMs) in diverse and challenging scenarios is essential to align them with human preferences. To mitigate the prohibitive costs associated with human evaluations, utilizing a powerful LLM as a judge has emerged as a favored approach. Nevertheless, this methodology encounters several challenges, including substantial expenses, concerns regarding privacy and security, and reproducibility. In this paper, we propose a straightforward, replicable, and accurate automated evaluation method by leveraging a lightweight LLM as the judge, named RocketEval. Initially, we identify that the performance disparity between lightweight and powerful LLMs in evaluation tasks primarily stems from their ability to conduct comprehensive analyses, which is not easily enhanced through techniques such as chain-of-thought reasoning. By reframing the evaluation task as a multi-faceted Q&A using an instance-specific checklist, we demonstrate that the limited judgment accuracy of lightweight LLMs is largely attributes to high uncertainty and positional bias. To address these challenges, we introduce an automated evaluation process grounded in checklist grading, which is designed to accommodate a variety of scenarios and questions. This process encompasses the creation of checklists, the grading of these checklists by lightweight LLMs, and the reweighting of checklist items to align with the supervised annotations. Our experiments carried out on the automated evaluation benchmarks, MT-Bench and WildBench datasets, reveal that RocketEval, when using Gemma-2-2B as the judge, achieves a high correlation (0.965) with human preferences, which is comparable to GPT-4o. Moreover, RocketEval provides a cost reduction exceeding 50-fold for large-scale evaluation and comparison scenarios. Our code is available at https://github.com/Joinn99/RocketEval-ICLR .
comment: Accepted by ICLR 2025: https://openreview.net/forum?id=zJjzNj6QUe
☆ Every FLOP Counts: Scaling a 300B Mixture-of-Experts LING LLM without Premium GPUs
In this technical report, we tackle the challenges of training large-scale Mixture of Experts (MoE) models, focusing on overcoming cost inefficiency and resource limitations prevalent in such systems. To address these issues, we present two differently sized MoE large language models (LLMs), namely Ling-Lite and Ling-Plus (referred to as "Bailing" in Chinese, spelled B\v{a}il\'ing in Pinyin). Ling-Lite contains 16.8 billion parameters with 2.75 billion activated parameters, while Ling-Plus boasts 290 billion parameters with 28.8 billion activated parameters. Both models exhibit comparable performance to leading industry benchmarks. This report offers actionable insights to improve the efficiency and accessibility of AI development in resource-constrained settings, promoting more scalable and sustainable technologies. Specifically, to reduce training costs for large-scale MoE models, we propose innovative methods for (1) optimization of model architecture and training processes, (2) refinement of training anomaly handling, and (3) enhancement of model evaluation efficiency. Additionally, leveraging high-quality data generated from knowledge graphs, our models demonstrate superior capabilities in tool use compared to other models. Ultimately, our experimental findings demonstrate that a 300B MoE LLM can be effectively trained on lower-performance devices while achieving comparable performance to models of a similar scale, including dense and MoE models. Compared to high-performance devices, utilizing a lower-specification hardware system during the pre-training phase demonstrates significant cost savings, reducing computing costs by approximately 20%. The models can be accessed at https://huggingface.co/inclusionAI.
comment: 34 pages
☆ AutoTestForge: A Multidimensional Automated Testing Framework for Natural Language Processing Models
In recent years, the application of behavioral testing in Natural Language Processing (NLP) model evaluation has experienced a remarkable and substantial growth. However, the existing methods continue to be restricted by the requirements for manual labor and the limited scope of capability assessment. To address these limitations, we introduce AutoTestForge, an automated and multidimensional testing framework for NLP models in this paper. Within AutoTestForge, through the utilization of Large Language Models (LLMs) to automatically generate test templates and instantiate them, manual involvement is significantly reduced. Additionally, a mechanism for the validation of test case labels based on differential testing is implemented which makes use of a multi-model voting system to guarantee the quality of test cases. The framework also extends the test suite across three dimensions, taxonomy, fairness, and robustness, offering a comprehensive evaluation of the capabilities of NLP models. This expansion enables a more in-depth and thorough assessment of the models, providing valuable insights into their strengths and weaknesses. A comprehensive evaluation across sentiment analysis (SA) and semantic textual similarity (STS) tasks demonstrates that AutoTestForge consistently outperforms existing datasets and testing tools, achieving higher error detection rates (an average of $30.89\%$ for SA and $34.58\%$ for STS). Moreover, different generation strategies exhibit stable effectiveness, with error detection rates ranging from $29.03\% - 36.82\%$.
comment: 15 pages, 4 figures, Under review
☆ SpecServe: Efficient and SLO-Aware Large Language Model Serving with Adaptive Speculative Decoding
Large Language Model (LLM) services often face challenges in achieving low inference latency and meeting Service Level Objectives (SLOs) under dynamic request patterns. Speculative decoding, which exploits lightweight models for drafting and LLMs for verification, has emerged as a compelling technique to accelerate LLM inference. However, existing speculative decoding solutions often fail to adapt to varying workloads and system environments, resulting in performance variability and SLO violations. In this paper, we introduce SpecServe, an efficient LLM inference system that dynamically adjusts speculative strategies according to real-time request loads and system configurations. SpecServe proposes a theoretical model to understand and predict the efficiency of speculative decoding across diverse scenarios. Additionally, it implements intelligent drafting and verification algorithms to guarantee optimal performance while achieving high SLO attainment. Experimental results on real-world LLM traces demonstrate that SpecServe consistently meets SLOs and achieves substantial performance improvements, yielding 1.14$\times$-14.3$\times$ speedups over state-of-the-art speculative inference systems.
☆ S2S-Arena, Evaluating Speech2Speech Protocols on Instruction Following with Paralinguistic Information
The rapid development of large language models (LLMs) has brought significant attention to speech models, particularly recent progress in speech2speech protocols supporting speech input and output. However, the existing benchmarks adopt automatic text-based evaluators for evaluating the instruction following ability of these models lack consideration for paralinguistic information in both speech understanding and generation. To address these issues, we introduce S2S-Arena, a novel arena-style S2S benchmark that evaluates instruction-following capabilities with paralinguistic information in both speech-in and speech-out across real-world tasks. We design 154 samples that fused TTS and live recordings in four domains with 21 tasks and manually evaluate existing popular speech models in an arena-style manner. The experimental results show that: (1) in addition to the superior performance of GPT-4o, the speech model of cascaded ASR, LLM, and TTS outperforms the jointly trained model after text-speech alignment in speech2speech protocols; (2) considering paralinguistic information, the knowledgeability of the speech model mainly depends on the LLM backbone, and the multilingual support of that is limited by the speech module; (3) excellent speech models can already understand the paralinguistic information in speech input, but generating appropriate audio with paralinguistic information is still a challenge.
☆ Capacity-Aware Inference: Mitigating the Straggler Effect in Mixture of Experts
The Mixture of Experts (MoE) is an effective architecture for scaling large language models by leveraging sparse expert activation, optimizing the trade-off between performance and efficiency. However, under expert parallelism, MoE suffers from inference inefficiencies due to imbalanced token-to-expert assignment, where some experts are overloaded while others remain underutilized. This imbalance leads to poor resource utilization and increased latency, as the most burdened expert dictates the overall delay, a phenomenon we define as the \textbf{\textit{Straggler Effect}}. To mitigate this, we propose Capacity-Aware Inference, including two key techniques: (1) \textbf{\textit{Capacity-Aware Token Drop}}, which discards overloaded tokens to regulate the maximum latency of MoE, and (2) \textbf{\textit{Capacity-Aware Token Reroute}}, which reallocates overflowed tokens to underutilized experts, balancing the token distribution. These techniques collectively optimize both high-load and low-load expert utilization, leading to a more efficient MoE inference pipeline. Extensive experiments demonstrate the effectiveness of our methods, showing significant improvements in inference efficiency, e.g., 0.2\% average performance increase and a 1.94$\times$ inference speedup on Mixtral-8$\times$7B-Instruct.
☆ The study of short texts in digital politics: Document aggregation for topic modeling
Statistical topic modeling is widely used in political science to study text. Researchers examine documents of varying lengths, from tweets to speeches. There is ongoing debate on how document length affects the interpretability of topic models. We investigate the effects of aggregating short documents into larger ones based on natural units that partition the corpus. In our study, we analyze one million tweets by U.S. state legislators from April 2016 to September 2020. We find that for documents aggregated at the account level, topics are more associated with individual states than when using individual tweets. This finding is replicated with Wikipedia pages aggregated by birth cities, showing how document definitions can impact topic modeling results.
☆ No Free Labels: Limitations of LLM-as-a-Judge Without Human Grounding
LLM-as-a-Judge is a framework that uses an LLM (large language model) to evaluate the quality of natural language text - typically text that is also generated by an LLM. This framework holds great promise due to its relative low-cost, ease of use, and strong correlations with human stylistic preferences. However, LLM Judges have been shown to exhibit biases that can distort their judgments. We evaluate how well LLM Judges can grade whether a given response to a conversational question is correct, an ability crucial to soundly estimating the overall response quality. To do so, we create and publicly release a human-annotated dataset with labels of correctness for 1,200 LLM responses. We source questions from a combination of existing datasets and a novel, challenging benchmark (BFF-Bench) created for this analysis. We demonstrate a strong connection between an LLM's ability to correctly answer a question and grade responses to that question. Although aggregate level statistics might imply a judge has high agreement with human annotators, it will struggle on the subset of questions it could not answer. To address this issue, we recommend a simple solution: provide the judge with a correct, human-written reference answer. We perform an in-depth analysis on how reference quality can affect the performance of an LLM Judge. We show that providing a weaker judge (e.g. Qwen 2.5 7B) with higher quality references reaches better agreement with human annotators than a stronger judge (e.g. GPT-4o) with synthetic references.
☆ ModernBERT is More Efficient than Conventional BERT for Chest CT Findings Classification in Japanese Radiology Reports
Objective: This study aims to evaluate and compare the performance of two Japanese language models-conventional Bidirectional Encoder Representations from Transformers (BERT) and the newer ModernBERT-in classifying findings from chest CT reports, with a focus on tokenization efficiency, processing time, and classification performance. Methods: We conducted a retrospective study using the CT-RATE-JPN dataset containing 22,778 training reports and 150 test reports. Both models were fine-tuned for multi-label classification of 18 common chest CT conditions. The training data was split in 18,222:4,556 for training and validation. Performance was evaluated using F1 scores for each condition and exact match accuracy across all 18 labels. Results: ModernBERT demonstrated superior tokenization efficiency, requiring 24.0% fewer tokens per document (258.1 vs. 339.6) compared to BERT Base. This translated to significant performance improvements, with ModernBERT completing training in 1877.67 seconds versus BERT's 3090.54 seconds (39% reduction). ModernBERT processed 38.82 samples per second during training (1.65x faster) and 139.90 samples per second during inference (1.66x faster). Despite these efficiency gains, classification performance remained comparable, with ModernBERT achieving superior F1 scores in 8 conditions, while BERT performed better in 4 conditions. Overall exact match accuracy was slightly higher for ModernBERT (74.67% vs. 72.67%), though this difference was not statistically significant (p=0.6291). Conclusion: ModernBERT offers substantial improvements in tokenization efficiency and training speed without sacrificing classification performance. These results suggest that ModernBERT is a promising candidate for clinical applications in Japanese radiology reports analysis.
comment: 23 pages, 8 figures
♻ ☆ Shifting Long-Context LLMs Research from Input to Output
Recent advancements in long-context Large Language Models (LLMs) have primarily concentrated on processing extended input contexts, resulting in significant strides in long-context comprehension. However, the equally critical aspect of generating long-form outputs has received comparatively less attention. This paper advocates for a paradigm shift in NLP research toward addressing the challenges of long-output generation. Tasks such as novel writing, long-term planning, and complex reasoning require models to understand extensive contexts and produce coherent, contextually rich, and logically consistent extended text. These demands highlight a critical gap in current LLM capabilities. We underscore the importance of this under-explored domain and call for focused efforts to develop foundational LLMs tailored for generating high-quality, long-form outputs, which hold immense potential for real-world applications.
comment: Preprint
♻ ☆ DIMSUM: Discourse in Mathematical Reasoning as a Supervision Module
We look at reasoning on GSM8k, a dataset of short texts presenting primary school, math problems. We find, with Mirzadeh et al. (2024), that current LLM progress on the data set may not be explained by better reasoning but by exposure to a broader pretraining data distribution. We then introduce a novel information source for helping models with less data or inferior training reason better: discourse structure. We show that discourse structure improves performance for models like Llama2 13b by up to 160%. Even for models that have most likely memorized the data set, adding discourse structural information to the model still improves predictions and dramatically improves large model performance on out of distribution examples.
♻ ☆ START: Self-taught Reasoner with Tools
Large reasoning models (LRMs) like OpenAI-o1 and DeepSeek-R1 have demonstrated remarkable capabilities in complex reasoning tasks through the utilization of long Chain-of-thought (CoT). However, these models often suffer from hallucinations and inefficiencies due to their reliance solely on internal reasoning processes. In this paper, we introduce START (Self-Taught Reasoner with Tools), a novel tool-integrated long CoT reasoning LLM that significantly enhances reasoning capabilities by leveraging external tools. Through code execution, START is capable of performing complex computations, self-checking, exploring diverse methods, and self-debugging, thereby addressing the limitations of LRMs. The core innovation of START lies in its self-learning framework, which comprises two key techniques: 1) Hint-infer: We demonstrate that inserting artificially designed hints (e.g., ``Wait, maybe using Python here is a good idea.'') during the inference process of a LRM effectively stimulates its ability to utilize external tools without the need for any demonstration data. Hint-infer can also serve as a simple and effective sequential test-time scaling method; 2) Hint Rejection Sampling Fine-Tuning (Hint-RFT): Hint-RFT combines Hint-infer and RFT by scoring, filtering, and modifying the reasoning trajectories with tool invocation generated by a LRM via Hint-infer, followed by fine-tuning the LRM. Through this framework, we have fine-tuned the QwQ-32B model to achieve START. On PhD-level science QA (GPQA), competition-level math benchmarks (AMC23, AIME24, AIME25), and the competition-level code benchmark (LiveCodeBench), START achieves accuracy rates of 63.6%, 95.0%, 66.7%, 47.1%, and 47.3%, respectively. It significantly outperforms the base QwQ-32B and achieves performance comparable to the state-of-the-art open-weight model R1-Distill-Qwen-32B and the proprietary model o1-Preview.
comment: 38 pages, 5 figures and 6 tables
♻ ☆ LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation
Assessing the reasoning capabilities of large language models (LLMs) is susceptible to overestimation due to data exposure of evaluation benchmarks. We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates and apply this framework to develop LINGOLY-TOO, a challenging benchmark for linguistic reasoning. By developing orthographic templates, we dynamically obfuscate the writing systems of real languages to generate numerousquestion variations. These variations preserve the reasoning steps required for each solution while reducing the likelihood of specific problem instances appearing in model training data. Our experiments demonstrate that frontier models, including Claud 3.7 Sonnet, o1-preview and DeepSeek R1, struggle with advanced reasoning. Our analysis also shows that LLMs exhibit noticeable variance in accuracy across permutations of the same problem, and on average perform better on questions appearing in their original orthography. Our findings highlight the opaque nature of response generation in LLMs and provide evidence that prior data exposure contributes to over estimating the reasoning capabilities of frontier models.
♻ ☆ Adding Alignment Control to Language Models
Post-training alignment has increasingly become a crucial factor in enhancing the usability of language models (LMs). However, the strength of alignment varies depending on individual preferences. This paper proposes a method to incorporate alignment control into a single model, referred to as CLM. This approach adds one identity layer preceding the initial layers and performs preference learning only on this layer to map unaligned input token embeddings into the aligned space. Experimental results demonstrate that this efficient fine-tuning method performs comparable to full fine-tuning. During inference, the input embeddings are processed through the aligned and unaligned layers, which are then merged through the interpolation coefficient. By controlling this parameter, the alignment exhibits a clear interpolation and extrapolation phenomenon.
♻ ☆ Ticktack : Long Span Temporal Alignment of Large Language Models Leveraging Sexagenary Cycle Time Expression
Large language models (LLMs) suffer from temporal misalignment issues especially across long span of time. The issue arises from knowing that LLMs are trained on large amounts of data where temporal information is rather sparse over long times, such as thousands of years, resulting in insufficient learning or catastrophic forgetting by the LLMs. This paper proposes a methodology named "Ticktack" for addressing the LLM's long-time span misalignment in a yearly setting. Specifically, we first propose to utilize the sexagenary year expression instead of the Gregorian year expression employed by LLMs, achieving a more uniform distribution in yearly granularity. Then, we employ polar coordinates to model the sexagenary cycle of 60 terms and the year order within each term, with additional temporal encoding to ensure LLMs understand them. Finally, we present a temporal representational alignment approach for post-training LLMs that effectively distinguishes time points with relevant knowledge, hence improving performance on time-related tasks, particularly over a long period. We also create a long time span benchmark for evaluation. Experimental results prove the effectiveness of our proposal.
♻ ☆ Chart-HQA: A Benchmark for Hypothetical Question Answering in Charts
Multimodal Large Language Models (MLLMs) have garnered significant attention for their strong visual-semantic understanding. Most existing chart benchmarks evaluate MLLMs' ability to parse information from charts to answer questions. However, they overlook the inherent output biases of MLLMs, where models rely on their parametric memory to answer questions rather than genuinely understanding the chart content. To address this limitation, we introduce a novel Chart Hypothetical Question Answering (HQA) task, which imposes assumptions on the same question to compel models to engage in counterfactual reasoning based on the chart content. Furthermore, we introduce HAI, a human-AI interactive data synthesis approach that leverages the efficient text-editing capabilities of LLMs alongside human expert knowledge to generate diverse and high-quality HQA data at a low cost. Using HAI, we construct Chart-HQA, a challenging benchmark synthesized from publicly available data sources. Evaluation results on 18 MLLMs of varying model sizes reveal that current models face significant generalization challenges and exhibit imbalanced reasoning performance on the HQA task.
comment: Under review
♻ ☆ Simple linear attention language models balance the recall-throughput tradeoff
Recent work has shown that attention-based language models excel at recall, the ability to ground generations in tokens previously seen in context. However, the efficiency of attention-based models is bottle-necked during inference by the KV-cache's aggressive memory consumption. In this work, we explore whether we can improve language model efficiency (e.g. by reducing memory consumption) without compromising on recall. By applying experiments and theory to a broad set of architectures, we identify a key tradeoff between a model's state size and recall ability. We show that efficient alternatives to attention (e.g. H3, Mamba, RWKV) maintain a fixed-size recurrent state, but struggle at recall. We propose BASED a simple architecture combining linear and sliding window attention. By varying BASED window size and linear attention feature dimension, we can dial the state size and traverse the pareto frontier of the recall-memory tradeoff curve, recovering the full quality of attention on one end and the small state size of attention-alternatives on the other. We train language models up to 1.3b parameters and show that BASED matches the strongest sub-quadratic models (e.g. Mamba) in perplexity and outperforms them on real-world recall-intensive tasks by 6.22 accuracy points. Implementations of linear attention are often less efficient than optimized standard attention implementations. To make BASED competitive, we develop IO-aware algorithms that enable 24x higher throughput on language generation than FlashAttention-2, when generating 1024 tokens using 1.3b parameter models. Code for this work is provided at: https://github.com/HazyResearch/based.
♻ ☆ Entangled Relations: Leveraging NLI and Meta-analysis to Enhance Biomedical Relation Extraction
Recent research efforts have explored the potential of leveraging natural language inference (NLI) techniques to enhance relation extraction (RE). In this vein, we introduce MetaEntailRE, a novel adaptation method that harnesses NLI principles to enhance RE performance. Our approach follows past works by verbalizing relation classes into class-indicative hypotheses, aligning a traditionally multi-class classification task to one of textual entailment. We introduce three key enhancements: (1) Meta-class analysis which, instead of labeling non-entailed premise-hypothesis pairs with the less informative "neutral" entailment label, provides additional context by analyzing overarching meta-relationships between classes; (2) Feasible hypothesis filtering, which removes unlikely hypotheses from consideration based on domain knowledge derived from data; and (3) Group-based prediction selection, which further improves performance by selecting highly confident predictions. MetaEntailRE is conceptually simple and empirically powerful, yielding significant improvements over conventional relation extraction techniques and other NLI formulations. We observe surprisingly large F1 gains of 17.6 points on BioRED and 13.4 points on ReTACRED compared to conventional methods, underscoring the versatility of MetaEntailRE across both biomedical and general domains.
comment: 17 pages, 1 figure
♻ ☆ DeltaProduct: Increasing the Expressivity of DeltaNet Through Products of Householders ICLR 2025
Linear Recurrent Neural Networks (linear RNNs) have emerged as competitive alternatives to Transformers for sequence modeling, offering efficient training and linear-time inference. However, existing architectures face a fundamental trade-off between expressivity and efficiency, dictated by the structure of their state-transition matrices. While diagonal matrices used in architectures like Mamba, GLA, or mLSTM yield fast runtime, they suffer from severely limited expressivity. To address this, recent architectures such as (Gated) DeltaNet and RWKVv7 adopted a diagonal plus rank-1 structure, allowing simultaneous token-channel mixing, which overcomes some expressivity limitations with only a slight decrease in training efficiency. Building on the interpretation of DeltaNet's recurrence as performing one step of online gradient descent per token on an associative recall loss, we introduce DeltaProduct, which instead takes multiple ($n_h$) steps per token. This naturally leads to diagonal plus rank-$n_h$ state-transition matrices, formed as products of $n_h$ generalized Householder transformations, providing a tunable mechanism to balance expressivity and efficiency and a stable recurrence. Through extensive experiments, we demonstrate that DeltaProduct achieves superior state-tracking and language modeling capabilities while exhibiting significantly improved length extrapolation compared to DeltaNet. Additionally, we also strengthen the theoretical foundation of DeltaNet's expressivity by proving that it can solve dihedral group word problems in just two layers.
comment: Accepted at ICLR 2025 Workshop on Foundation Models in the Wild
♻ ☆ DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inference ICLR'25
Large language models (LLMs) are increasingly employed for complex tasks that process multiple generation calls in a tree structure with shared prefixes of tokens, including few-shot prompting, multi-step reasoning, speculative decoding, etc. However, existing inference systems for tree-based applications are inefficient due to improper partitioning of queries and KV cache during attention calculation. This leads to two main issues: (1) a lack of memory access (IO) reuse for KV cache of shared prefixes, and (2) poor load balancing.As a result, there is redundant KV cache IO between GPU global memory and shared memory, along with low GPU utilization. To address these challenges, we propose DeFT(Decoding with Flash Tree-Attention), a hardware-efficient attention algorithm with prefix-aware and load-balanced KV cache partitions. DeFT reduces the number of read/write operations of KV cache during attention calculation through KV-Guided Grouping, a method that avoids repeatedly loading KV cache of shared prefixes in attention computation. Additionally, we propose Flattened Tree KV Splitting, a mechanism that ensures even distribution of the KV cache across partitions with little computation redundancy, enhancing GPU utilization during attention computations. By reducing 73-99% KV cache IO and nearly 100% IO for partial results during attention calculation, DeFT achieves up to 2.23/3.59x speedup in the end-to-end/attention latency across three practical tree-based workloads compared to state-of-the-art attention algorithms. Our code is available at https://github.com/LINs-lab/DeFT.
comment: Update DeFT-v4, accepted by ICLR'25 (https://openreview.net/forum?id=2c7pfOqu9k). Our code is available at https://github.com/LINs-lab/DeFT
♻ ☆ Language Models Enable Simple Systems for Generating Structured Views of Heterogeneous Data Lakes
A long standing goal of the data management community is to develop general, automated systems that ingest semi-structured documents and output queryable tables without human effort or domain specific customization. Given the sheer variety of potential documents, state-of-the art systems make simplifying assumptions and use domain specific training. In this work, we ask whether we can maintain generality by using large language models (LLMs). LLMs, which are pretrained on broad data, can perform diverse downstream tasks simply conditioned on natural language task descriptions. We propose and evaluate EVAPORATE, a simple, prototype system powered by LLMs. We identify two fundamentally different strategies for implementing this system: prompt the LLM to directly extract values from documents or prompt the LLM to synthesize code that performs the extraction. Our evaluations show a cost-quality tradeoff between these two approaches. Code synthesis is cheap, but far less accurate than directly processing each document with the LLM. To improve quality while maintaining low cost, we propose an extended code synthesis implementation, EVAPORATE-CODE+, which achieves better quality than direct extraction. Our key insight is to generate many candidate functions and ensemble their extractions using weak supervision. EVAPORATE-CODE+ not only outperforms the state-of-the art systems, but does so using a sublinear pass over the documents with the LLM. This equates to a 110x reduction in the number of tokens the LLM needs to process, averaged across 16 real-world evaluation settings of 10k documents each.
♻ ☆ Correcting Annotator Bias in Training Data: Population-Aligned Instance Replication (PAIR)
Models trained on crowdsourced labels may not reflect broader population views, because those who work as annotators do not represent the population. We propose Population-Aligned Instance Replication (PAIR), a method to address bias caused by non-representative annotator pools. Using a simulation study of offensive language and hate speech, we create two types of annotators with different labeling tendencies and generate datasets with varying proportions of the types. We observe that models trained on unbalanced annotator pools show poor calibration compared to those trained on representative data. By duplicating labels from underrepresented annotator groups to match population proportions, PAIR reduces bias without collecting additional annotations. These results suggest that statistical techniques from survey research can improve model performance. We conclude with practical recommendations for improving the representativity of training data and model performance.
♻ ☆ SynSUM -- Synthetic Benchmark with Structured and Unstructured Medical Records AAAI 2025
We present the SynSUM benchmark, a synthetic dataset linking unstructured clinical notes to structured background variables. The dataset consists of 10,000 artificial patient records containing tabular variables (like symptoms, diagnoses and underlying conditions) and related notes describing the fictional patient encounter in the domain of respiratory diseases. The tabular portion of the data is generated through a Bayesian network, where both the causal structure between the variables and the conditional probabilities are proposed by an expert based on domain knowledge. We then prompt a large language model (GPT-4o) to generate a clinical note related to this patient encounter, describing the patient symptoms and additional context. We conduct both an expert evaluation study to assess the quality of the generated notes, as well as running some simple predictor models on both the tabular and text portions of the dataset, forming a baseline for further research. The SynSUM dataset is primarily designed to facilitate research on clinical information extraction in the presence of tabular background variables, which can be linked through domain knowledge to concepts of interest to be extracted from the text - the symptoms, in the case of SynSUM. Secondary uses include research on the automation of clinical reasoning over both tabular data and text, causal effect estimation in the presence of tabular and/or textual confounders, and multi-modal synthetic data generation.
comment: The dataset can be downloaded from https://github.com/prabaey/synsum. Presented at the GenAI4Health workshop at AAAI 2025
♻ ☆ AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models
Large language models (LLMs) often exhibit hallucinations due to incorrect or outdated knowledge. Hence, model editing methods have emerged to enable targeted knowledge updates. To achieve this, a prevailing paradigm is the locating-then-editing approach, which first locates influential parameters and then edits them by introducing a perturbation. While effective, current studies have demonstrated that this perturbation inevitably disrupt the originally preserved knowledge within LLMs, especially in sequential editing scenarios. To address this, we introduce AlphaEdit, a novel solution that projects perturbation onto the null space of the preserved knowledge before applying it to the parameters. We theoretically prove that this projection ensures the output of post-edited LLMs remains unchanged when queried about the preserved knowledge, thereby mitigating the issue of disruption. Extensive experiments on various LLMs, including LLaMA3, GPT2-XL, and GPT-J, show that AlphaEdit boosts the performance of most locating-then-editing methods by an average of 36.4% with a single line of additional code for projection solely. Our code is available at: https://github.com/jianghoucheng/AlphaEdit.
♻ ☆ The interplay between domain specialization and model size
Scaling laws for language models have often focused on finding the optimal model size and token count for training from scratch. However, achieving this optimal balance requires significant compute resources due to the extensive data demands when training models from randomly-initialized weights. Continued pretraining offers a cost-effective alternative, leveraging the compute investment from pretrained models to incorporate new knowledge without requiring extensive new data. Recent findings suggest that data quality influences constants in scaling laws, thereby altering the optimal parameter-token allocation ratio. Building on this insight, we investigate the interplay between domain specialization and model size during continued pretraining under compute-constrained scenarios. Our goal is to identify an optimal training regime for this scenario and detect patterns in this interplay that can be generalized across different model sizes and domains. To compare general and specialized training, we filtered a web-based dataset to extract data from three domains: legal, medical, and accounting. We pretrained models with 1.5B, 3B, 7B, and 14B parameters on both the unfiltered and filtered datasets, then evaluated their performance on domain-specific exams. Results show that as model size increases, specialized models outperform general models while requiring less training compute. Additionally, their growing compute efficiency leads to reduced forgetting of previously learned knowledge.
♻ ☆ SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)
Whether LLMs memorize their training data and what this means, from measuring privacy leakage to detecting copyright violations, has become a rapidly growing area of research. In the last few months, more than 10 new methods have been proposed to perform Membership Inference Attacks (MIAs) against LLMs. Contrary to traditional MIAs which rely on fixed-but randomized-records or models, these methods are mostly trained and tested on datasets collected post-hoc. Sets of members and non-members, used to evaluate the MIA, are constructed using informed guesses after the release of a model. This lack of randomization raises concerns of a distribution shift between members and non-members. In this work, we first extensively review the literature on MIAs against LLMs and show that, while most work focuses on sequence-level MIAs evaluated in post-hoc setups, a range of target models, motivations and units of interest are considered. We then quantify distribution shifts present in 6 datasets used in the literature using a model-less bag of word classifier and show that all datasets constructed post-hoc suffer from strong distribution shifts. These shifts invalidate the claims of LLMs memorizing strongly in real-world scenarios and, potentially, also the methodological contributions of the recent papers based on these datasets. Yet, all hope might not be lost. We introduce important considerations to properly evaluate MIAs against LLMs and discuss, in turn, potential ways forwards: randomized test splits, injections of randomized (unique) sequences, randomized fine-tuning, and several post-hoc control methods. While each option comes with its advantages and limitations, we believe they collectively provide solid grounds to guide MIA development and study LLM memorization. We conclude with an overview of recommended approaches to benchmark sequence-level and document-level MIAs against LLMs.
comment: IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2025)
♻ ☆ LiGT: Layout-infused Generative Transformer for Visual Question Answering on Vietnamese Receipts
Document Visual Question Answering (Document VQA) challenges multimodal systems to holistically handle textual, layout, and visual modalities to provide appropriate answers. Document VQA has gained popularity in recent years due to the increasing amount of documents and the high demand for digitization. Nonetheless, most of document VQA datasets are developed in high-resource languages such as English. In this paper, we present ReceiptVQA (\textbf{Receipt} \textbf{V}isual \textbf{Q}uestion \textbf{A}nswering), the initial large-scale document VQA dataset in Vietnamese dedicated to receipts, a document kind with high commercial potentials. The dataset encompasses \textbf{9,000+} receipt images and \textbf{60,000+} manually annotated question-answer pairs. In addition to our study, we introduce LiGT (\textbf{L}ayout-\textbf{i}nfused \textbf{G}enerative \textbf{T}ransformer), a layout-aware encoder-decoder architecture designed to leverage embedding layers of language models to operate layout embeddings, minimizing the use of additional neural modules. Experiments on ReceiptVQA show that our architecture yielded promising performance, achieving competitive results compared with outstanding baselines. Furthermore, throughout analyzing experimental results, we found evident patterns that employing encoder-only model architectures has considerable disadvantages in comparison to architectures that can generate answers. We also observed that it is necessary to combine multiple modalities to tackle our dataset, despite the critical role of semantic understanding from language models. We hope that our work will encourage and facilitate future development in Vietnamese document VQA, contributing to a diverse multimodal research community in the Vietnamese language.
comment: Accepted at IJDAR
♻ ☆ Bootstrapping Language Models with DPO Implicit Rewards ICLR 2025
Human alignment in large language models (LLMs) is an active area of research. A recent groundbreaking work, direct preference optimization (DPO), has greatly simplified the process from past work in reinforcement learning from human feedback (RLHF) by bypassing the reward learning stage in RLHF. DPO, after training, provides an implicit reward model. In this work, we make a novel observation that this implicit reward model can by itself be used in a bootstrapping fashion to further align the LLM. Our approach is to use the rewards from a current LLM to construct a preference dataset, which is then used in subsequent DPO rounds. We incorporate two refinements to further improve our approach: 1) length-regularized reward shaping to make the preference dataset length-unbiased; 2) experience replay to enhance the quality of the preference dataset. Our approach, named self-alignment with DPO ImpliCit rEwards (DICE), shows great improvements in alignment. It achieves an increase of more than 8$\\%$ in lengthcontrolled win rate on AlpacaEval 2 for all the different base models that we tried, without relying on external feedback. Our code is available at https://github.com/sail-sg/dice.
comment: Accepted in ICLR 2025
♻ ☆ NLI under the Microscope: What Atomic Hypothesis Decomposition Reveals NAACL 2025
Decomposition of text into atomic propositions is a flexible framework allowing for the closer inspection of input and output text. We use atomic decomposition of hypotheses in two natural language reasoning tasks, traditional NLI and defeasible NLI, to form atomic sub-problems, or granular inferences that models must weigh when solving the overall problem. These atomic sub-problems serve as a tool to further understand the structure of both NLI and defeasible reasoning, probe a model's consistency and understanding of different inferences, and measure the diversity of examples in benchmark datasets. Our results indicate that LLMs still struggle with logical consistency on atomic NLI and defeasible NLI sub-problems. Lastly, we identify critical atomic sub-problems of defeasible NLI examples, or those that most contribute to the overall label, and propose a method to measure the inferential consistency of a model, a metric designed to capture the degree to which a model makes consistently correct or incorrect predictions about the same fact under different contexts.
comment: Accepted to NAACL 2025
♻ ☆ CNsum:Automatic Summarization for Chinese News Text
Obtaining valuable information from massive data efficiently has become our research goal in the era of Big Data. Text summarization technology has been continuously developed to meet this demand. Recent work has also shown that transformer-based pre-trained language models have achieved great success on various tasks in Natural Language Processing (NLP). Aiming at the problem of Chinese news text summary generation and the application of Transformer structure on Chinese, this paper proposes a Chinese news text summarization model (CNsum) based on Transformer structure, and tests it on Chinese datasets such as THUCNews. The results of the conducted experiments show that CNsum achieves better ROUGE score than the baseline models, which verifies the outperformance of the model.
comment: This withdrawal is due to the lack of authorization from all co-authors for the publication of this version
♻ ☆ MeanCache: User-Centric Semantic Caching for LLM Web Services
Large Language Models (LLMs) like ChatGPT and Llama have revolutionized natural language processing and search engine dynamics. However, these models incur exceptionally high computational costs. For instance, GPT-3 consists of 175 billion parameters, where inference demands billions of floating-point operations. Caching is a natural solution to reduce LLM inference costs on repeated queries, which constitute about 31% of the total queries. However, existing caching methods are incapable of finding semantic similarities among LLM queries nor do they operate on contextual queries, leading to unacceptable false hit-and-miss rates. This paper introduces MeanCache, a user-centric semantic cache for LLM-based services that identifies semantically similar queries to determine cache hit or miss. Using MeanCache, the response to a user's semantically similar query can be retrieved from a local cache rather than re-querying the LLM, thus reducing costs, service provider load, and environmental impact. MeanCache leverages Federated Learning (FL) to collaboratively train a query similarity model without violating user privacy. By placing a local cache in each user's device and using FL, MeanCache reduces the latency and costs and enhances model performance, resulting in lower false hit rates. MeanCache also encodes context chains for every cached query, offering a simple yet highly effective mechanism to discern contextual query responses from standalone. Our experiments benchmarked against the state-of-the-art caching method, reveal that MeanCache attains an approximately 17% higher F-score and a 20% increase in precision during semantic cache hit-and-miss decisions while performing even better on contextual queries. It also reduces the storage requirement by 83% and accelerates semantic cache hit-and-miss decisions by 11%.
comment: Accepted at 2025 IEEE 39th International Parallel and Distributed Processing Symposium (IPDPS)
♻ ☆ AILS-NTUA at SemEval-2025 Task 8: Language-to-Code prompting and Error Fixing for Tabular Question Answering
In this paper, we present our submission to SemEval-2025 Task 8: Question Answering over Tabular Data. This task, evaluated on the DataBench dataset, assesses Large Language Models' (LLMs) ability to answer natural language questions over structured data while addressing topic diversity and table size limitations in previous benchmarks. We propose a system that employs effective LLM prompting to translate natural language queries into executable code, enabling accurate responses, error correction, and interpretability. Our approach ranks first in both subtasks of the competition in the proprietary model category, significantly outperforming the organizer's baseline.
♻ ☆ LIFT: Improving Long Context Understanding of Large Language Models through Long Input Fine-Tuning
Long context understanding remains challenging for large language models due to their limited context windows. This paper presents Long Input Fine-Tuning (LIFT), a novel framework for long-context modeling that can improve the long-context performance of arbitrary (short-context) LLMs by dynamically adapting model parameters based on the long input. Importantly, LIFT, rather than endlessly extending the context window size to accommodate increasingly longer inputs in context, chooses to store and absorb the long input in parameter. By fine-tuning the long input into model parameters, LIFT allows short-context LLMs to answer questions even when the required information is not provided in the context during inference. Furthermore, to enhance LIFT performance while maintaining the original in-context learning (ICL) capabilities, we introduce Gated Memory, a specialized attention adapter that automatically balances long input memorization and ICL. We provide a comprehensive analysis of the strengths and limitations of LIFT on long context understanding, offering valuable directions for future research.
comment: arXiv admin note: text overlap with arXiv:2412.13626
♻ ☆ ECCOS: Efficient Capability and Cost Coordinated Scheduling for Multi-LLM Serving
As large language models (LLMs) are increasingly deployed as service endpoints in systems, the surge in query volume creates significant scheduling challenges. Existing scheduling frameworks mainly target at latency optimization while neglecting the capability of LLMs to serve different level of queries, which could lead to computational resource waste. This paper addresses this challenge by proposing a capability-cost coordinated scheduling framework, ECCOS, for multi-LLM serving, which explicitly constrains response quality and workload to optimize LLM inference cost. Specifically, it introduces the two-stage scheduling by designing a multi-objective predictor and a constrained optimizer. The predictor estimates both model capabilities and computational costs through training-based and retrieval-based approaches, while the optimizer determines cost-optimal assignments under quality and workload constraints. It also introduces QAServe, a dataset collected for sample-wise response quality and costs by zero-shot prompting different LLMs on knowledge QA and mathematical reasoning. Extensive experiments demonstrate that ECCOS improves success rates by 6.30% while reducing costs by 10.15% compared to existing methods, consuming less than 0.5% of LLM response time. The code is available at: https://github.com/agiresearch/ECCOS.
♻ ☆ Emergent Language: A Survey and Taxonomy
The field of emergent language represents a novel area of research within the domain of artificial intelligence, particularly within the context of multi-agent reinforcement learning. Although the concept of studying language emergence is not new, early approaches were primarily concerned with explaining human language formation, with little consideration given to its potential utility for artificial agents. In contrast, studies based on reinforcement learning aim to develop communicative capabilities in agents that are comparable to or even superior to human language. Thus, they extend beyond the learned statistical representations that are common in natural language processing research. This gives rise to a number of fundamental questions, from the prerequisites for language emergence to the criteria for measuring its success. This paper addresses these questions by providing a comprehensive review of 181 scientific publications on emergent language in artificial intelligence. Its objective is to serve as a reference for researchers interested in or proficient in the field. Consequently, the main contributions are the definition and overview of the prevailing terminology, the analysis of existing evaluation methods and metrics, and the description of the identified research gaps.
comment: published in Journal of Autonomous Agents and Multi-Agent Systems
♻ ☆ Familiarity: Better Evaluation of Zero-Shot Named Entity Recognition by Quantifying Label Shifts in Synthetic Training Data
Zero-shot named entity recognition (NER) is the task of detecting named entities of specific types (such as 'Person' or 'Medicine') without any training examples. Current research increasingly relies on large synthetic datasets, automatically generated to cover tens of thousands of distinct entity types, to train zero-shot NER models. However, in this paper, we find that these synthetic datasets often contain entity types that are semantically highly similar to (or even the same as) those in standard evaluation benchmarks. Because of this overlap, we argue that reported F1 scores for zero-shot NER overestimate the true capabilities of these approaches. Further, we argue that current evaluation setups provide an incomplete picture of zero-shot abilities since they do not quantify the label shift (i.e., the similarity of labels) between training and evaluation datasets. To address these issues, we propose Familiarity, a novel metric that captures both the semantic similarity between entity types in training and evaluation, as well as their frequency in the training data, to provide an estimate of label shift. It allows researchers to contextualize reported zero-shot NER scores when using custom synthetic training datasets. Further, it enables researchers to generate evaluation setups of various transfer difficulties for fine-grained analysis of zero-shot NER.
comment: 9 pages, 4 figures, 5 tables
♻ ☆ A Confidence-based Acquisition Model for Self-supervised Active Learning and Label Correction
Supervised neural approaches are hindered by their dependence on large, meticulously annotated datasets, a requirement that is particularly cumbersome for sequential tasks. The quality of annotations tends to deteriorate with the transition from expert-based to crowd-sourced labelling. To address these challenges, we present CAMEL (Confidence-based Acquisition Model for Efficient self-supervised active Learning), a pool-based active learning framework tailored to sequential multi-output problems. CAMEL possesses two core features: (1) it requires expert annotators to label only a fraction of a chosen sequence, and (2) it facilitates self-supervision for the remainder of the sequence. By deploying a label correction mechanism, CAMEL can also be utilised for data cleaning. We evaluate CAMEL on two sequential tasks, with a special emphasis on dialogue belief tracking, a task plagued by the constraints of limited and noisy datasets. Our experiments demonstrate that CAMEL significantly outperforms the baselines in terms of efficiency. Furthermore, the data corrections suggested by our method contribute to an overall improvement in the quality of the resulting datasets.
♻ ☆ EdgeMoE: Empowering Sparse Large Language Models on Mobile Devices
Large language models (LLMs) such as GPTs and Mixtral-8x7B have revolutionized machine intelligence due to their exceptional abilities in generic ML tasks. Transiting LLMs from datacenters to edge devices brings benefits like better privacy and availability, but is challenged by their massive parameter size and thus unbearable runtime costs. To this end, we present EdgeMoE, an on-device inference engine for mixture-of-expert (MoE) LLMs -- a popular form of sparse LLM that scales its parameter size with almost constant computing complexity. EdgeMoE achieves both memory- and compute-efficiency by partitioning the model into the storage hierarchy: non-expert weights are held in device memory; while expert weights are held on external storage and fetched to memory only when activated. This design is motivated by a key observation that expert weights are bulky but infrequently used due to sparse activation. To further reduce the expert I/O swapping overhead, EdgeMoE incorporates two novel techniques: (1) expert-wise bitwidth adaptation that reduces the expert sizes with tolerable accuracy loss; (2) expert preloading that predicts the activated experts ahead of time and preloads it with the compute-I/O pipeline. On popular MoE LLMs and edge devices, EdgeMoE showcase significant memory savings and speedup over competitive baselines. The code is available at https://github.com/UbiquitousLearning/mllm.
♻ ☆ Dialogue Ontology Relation Extraction via Constrained Chain-of-Thought Decoding SIGDIAL 2024
State-of-the-art task-oriented dialogue systems typically rely on task-specific ontologies for fulfilling user queries. The majority of task-oriented dialogue data, such as customer service recordings, comes without ontology and annotation. Such ontologies are normally built manually, limiting the application of specialised systems. Dialogue ontology construction is an approach for automating that process and typically consists of two steps: term extraction and relation extraction. In this work, we focus on relation extraction in a transfer learning set-up. To improve the generalisation, we propose an extension to the decoding mechanism of large language models. We adapt Chain-of-Thought (CoT) decoding, recently developed for reasoning problems, to generative relation extraction. Here, we generate multiple branches in the decoding space and select the relations based on a confidence threshold. By constraining the decoding to ontology terms and relations, we aim to decrease the risk of hallucination. We conduct extensive experimentation on two widely used datasets and find improvements in performance on target ontology for source fine-tuned and one-shot prompted large language models.
comment: Accepted to appear at SIGDIAL 2024. 9 pages, 4 figures
♻ ☆ LongEval: A Comprehensive Analysis of Long-Text Generation Through a Plan-based Paradigm
Large Language Models (LLMs) have achieved remarkable success in various natural language processing tasks, yet their ability to generate long-form content remains poorly understood and evaluated. Our analysis reveals that current LLMs struggle with length requirements and information density in long-text generation, with performance deteriorating as text length increases. To quantitively locate such a performance degradation and provide further insights on model development, we present LongEval, a benchmark that evaluates long-text generation through both direct and plan-based generation paradigms, inspired by cognitive and linguistic writing models. The comprehensive experiments in this work reveal interesting findings such as that while model size correlates with generation ability, the small-scale model (e.g., LongWriter), well-trained on long texts, has comparable performance. All code and datasets are released in https://github.com/Wusiwei0410/LongEval.
comment: Under review
♻ ☆ Are AI Detectors Good Enough? A Survey on Quality of Datasets With Machine-Generated Texts AAAI 2025
The rapid development of autoregressive Large Language Models (LLMs) has significantly improved the quality of generated texts, necessitating reliable machine-generated text detectors. A huge number of detectors and collections with AI fragments have emerged, and several detection methods even showed recognition quality up to 99.9% according to the target metrics in such collections. However, the quality of such detectors tends to drop dramatically in the wild, posing a question: Are detectors actually highly trustworthy or do their high benchmark scores come from the poor quality of evaluation datasets? In this paper, we emphasise the need for robust and qualitative methods for evaluating generated data to be secure against bias and low generalising ability of future model. We present a systematic review of datasets from competitions dedicated to AI-generated content detection and propose methods for evaluating the quality of datasets containing AI-generated fragments. In addition, we discuss the possibility of using high-quality generated data to achieve two goals: improving the training of detection models and improving the training datasets themselves. Our contribution aims to facilitate a better understanding of the dynamics between human and machine text, which will ultimately support the integrity of information in an increasingly automated world. The code is available at https://github.com/Advacheck-OU/ai-dataset-analysing.
comment: Presented at Preventing and Detecting LLM Misinformation (PDLM) at AAAI 2025
♻ ☆ RoToR: Towards More Reliable Responses for Order-Invariant Inputs
Mitigating positional bias of language models (LMs) for listwise inputs is a well-known and important problem (e.g., lost-in-the-middle). While zero-shot order-invariant LMs have been proposed to solve this issue, their success on practical listwise problems has been limited. In this work, as a first contribution, we identify and overcome two limitations to make zero-shot invariant LMs more practical: (1) training and inference distribution mismatch arising from modifying positional ID assignments to enforce invariance, and (2) failure to adapt to a mixture of order-invariant and sensitive inputs in practical listwise problems. Then, to overcome these issues we propose (1) RoToR, a zero-shot invariant LM for genuinely order-invariant inputs with minimal modifications of positional IDs, and (2) Selective Routing, an adaptive framework that handles both order-invariant and order-sensitive inputs in listwise tasks. On the Lost in the middle (LitM), Knowledge Graph QA (KGQA), and MMLU benchmarks, we show that RoToR with Selective Routing can effectively handle practical listwise input tasks in a zero-shot manner.
♻ ☆ DetectRL: Benchmarking LLM-Generated Text Detection in Real-World Scenarios NeurIPS 2024
Detecting text generated by large language models (LLMs) is of great recent interest. With zero-shot methods like DetectGPT, detection capabilities have reached impressive levels. However, the reliability of existing detectors in real-world applications remains underexplored. In this study, we present a new benchmark, DetectRL, highlighting that even state-of-the-art (SOTA) detection techniques still underperformed in this task. We collected human-written datasets from domains where LLMs are particularly prone to misuse. Using popular LLMs, we generated data that better aligns with real-world applications. Unlike previous studies, we employed heuristic rules to create adversarial LLM-generated text, simulating various prompts usages, human revisions like word substitutions, and writing noises like spelling mistakes. Our development of DetectRL reveals the strengths and limitations of current SOTA detectors. More importantly, we analyzed the potential impact of writing styles, model types, attack methods, the text lengths, and real-world human writing factors on different types of detectors. We believe DetectRL could serve as an effective benchmark for assessing detectors in real-world scenarios, evolving with advanced attack methods, thus providing more stressful evaluation to drive the development of more efficient detectors. Data and code are publicly available at: https://github.com/NLP2CT/DetectRL.
comment: Accepted to NeurIPS 2024 Datasets and Benchmarks Track (Camera-Ready)
♻ ☆ Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-quality web and synthetic data, significantly outperforming recent open-source models of similar size and matching the performance of models twice its size on math and coding tasks requiring complex reasoning. This achievement is driven by a carefully curated synthetic data recipe emphasizing high-quality math and coding datasets. Compared to its predecessor, Phi-3.5-Mini, Phi-4-Mini features an expanded vocabulary size of 200K tokens to better support multilingual applications, as well as group query attention for more efficient long-sequence generation. Phi-4-Multimodal is a multimodal model that integrates text, vision, and speech/audio input modalities into a single model. Its novel modality extension approach leverages LoRA adapters and modality-specific routers to allow multiple inference modes combining various modalities without interference. For example, it now ranks first in the OpenASR leaderboard to date, although the LoRA component of the speech/audio modality has just 460 million parameters. Phi-4-Multimodal supports scenarios involving (vision + language), (vision + speech), and (speech/audio) inputs, outperforming larger vision-language and speech-language models on a wide range of tasks. Additionally, we experiment to further train Phi-4-Mini to enhance its reasoning capabilities. Despite its compact 3.8-billion-parameter size, this experimental version achieves reasoning performance on par with or surpassing significantly larger models, including DeepSeek-R1-Distill-Qwen-7B and DeepSeek-R1-Distill-Llama-8B.
comment: 39 pages
♻ ☆ AdEval: Alignment-based Dynamic Evaluation to Mitigate Data Contamination in Large Language Models
As Large Language Models (LLMs) are pretrained on massive-scale corpora, the issue of data contamination has become increasingly severe, leading to potential overestimation of model performance during evaluation. To address this, we propose AdEval (Alignment-based Dynamic Evaluation), a dynamic data evaluation method aimed at mitigating the impact of data contamination on evaluation reliability. Experimental results on multiple datasets demonstrate that AdEval effectively reduces the impact of data contamination on evaluation outcomes, enhancing both the fairness and reliability of the evaluation process.
comment: There are serious academic problems in this paper, such as data falsification and plagiarism in the method of the paper
♻ ☆ Transformers for molecular property prediction: Domain adaptation efficiently improves performance
Most of the current transformer-based chemical language models are pre-trained on millions to billions of molecules. However, the improvement from such scaling in dataset size is not confidently linked to improved molecular property prediction. The aim of this study is to investigate and overcome some of the limitations of transformer models in predicting molecular properties. Specifically, we examine the impact of pre-training dataset size and diversity on the performance of transformer models and investigate the use of domain adaptation as a technique for improving model performance. First, our findings indicate that increasing pretraining dataset size beyond 400K molecules from the GuacaMol dataset does not result in a significant improvement on four ADME endpoints, namely, solubility, permeability, microsomal stability, and plasma protein binding. Second, our results demonstrate that using domain adaptation by further training the transformer model on a small set of domain-relevant molecules, i.e., a few hundred to a few thousand, using multi-task regression of physicochemical properties was sufficient to significantly improve performance for three out of the four investigated ADME endpoints (P-value < 0.001). Finally, we observe that a model pre-trained on 400K molecules and domain adopted on a few hundred/thousand molecules performs similarly (P-value > 0.05) to more complicated transformer models like MolBERT(pre-trained on 1.3M molecules) and MolFormer (pre-trained on 100M molecules). A comparison to a random forest model trained on basic physicochemical properties showed similar performance to the examined transformer models. We believe that current transformer models can be improved through further systematic analysis of pre-training and downstream data, pre-training objectives, and scaling laws, ultimately leading to better and more helpful models.
♻ ☆ ARIES: Stimulating Self-Refinement of Large Language Models by Iterative Preference Optimization
A truly intelligent Large Language Model (LLM) should be capable of correcting errors in its responses through external interactions. However, even the most advanced models often face challenges in improving their outputs. In this paper, we explore how to cultivate LLMs with the self-refinement capability through iterative preference training, and how this ability can be leveraged to improve model performance during inference. To this end, we introduce a novel post-training and inference framework, called ARIES: Adaptive Refinement and Iterative Enhancement Structure. This method iteratively performs preference training and self-refinement-based data collection. During training, ARIES strengthen the model's direct question-answering capability while simultaneously unlocking its self-refinement potential. During inference, ARIES harnesses this self-refinement capability to generate a series of progressively refined responses, which are then filtered using either the Reward Model Scoring or a simple yet effective Rule-Based Selection mechanism, specifically tailored to our approach, to construct a dataset for the next round of preference training. Experimental results demonstrate the remarkable performance of ARIES. When applied to the Llama-3.1-8B model and under the self-refinement setting, ARIES surpasses powerful models such as GPT-4o, achieving 62.3% length-controlled (LC) and a 63.3% raw win rates on AlpacaEval 2, outperforming Iterative DPO by 27.8% and 35.5% respectively, as well as a 50.3% win rate on Arena-Hard, surpassing Iterative DPO by 26.6%. Furthermore, ARIES consistently enhances performance on mathematical reasoning tasks like GSM8K and MATH.
♻ ☆ CLIP meets DINO for Tuning Zero-Shot Classifier using Unlabeled Image Collections
In the era of foundation models, CLIP has emerged as a powerful tool for aligning text & visual modalities into a common embedding space. However, the alignment objective used to train CLIP often results in subpar visual features for fine-grained tasks. In contrast, SSL-pretrained models like DINO excel at extracting rich visual features due to their specialized training paradigm. Yet, these SSL models require an additional supervised linear probing step, which relies on fully labeled data which is often expensive and difficult to obtain at scale. In this paper, we propose a label-free prompt-tuning method that leverages the rich visual features of self-supervised learning models (DINO) and the broad textual knowledge of large language models (LLMs) to largely enhance CLIP-based image classification performance using unlabeled images. Our approach unfolds in three key steps: (1) We generate robust textual feature embeddings that more accurately represent object classes by leveraging class-specific descriptions from LLMs, enabling more effective zero-shot classification compared to CLIP's default name-specific prompts. (2) These textual embeddings are then used to produce pseudo-labels to train an alignment module that integrates the complementary strengths of LLM description-based textual embeddings & DINO's visual features. (3) Finally, we prompt-tune CLIP's vision encoder through DINO-assisted supervision using the trained alignment module. This three-step process allows us to harness the best of visual & textual foundation models, resulting in a powerful and efficient approach that surpasses state-of-the-art label-free classification methods. Notably, our framework, NoLA (No Labels Attached), achieves an average absolute gain of 3.6% over the state-of-the-art LaFTer across 11 diverse image classification datasets. Our code & models can be found at https://github.com/fazliimam/NoLA.
♻ ☆ LLM-based Discriminative Reasoning for Knowledge Graph Question Answering
Large language models (LLMs) based on generative pre-trained Transformer have achieved remarkable performance on knowledge graph question-answering (KGQA) tasks. However, LLMs often produce ungrounded subgraph planning or reasoning results in KGQA due to the hallucinatory behavior brought by the generative paradigm. To tackle this issue, we propose READS to reformulate the KGQA process into discriminative subtasks, which simplifies the search space for each subtasks. Based on the subtasks, we design a new corresponding discriminative inference strategy to conduct the reasoning for KGQA, thereby alleviating hallucination and ungrounded reasoning issues in LLMs. Experimental results show that the proposed approach outperforms multiple strong comparison methods, along with achieving state-of-the-art performance on widely used benchmarks WebQSP and CWQ.
♻ ☆ Explicit vs. Implicit: Investigating Social Bias in Large Language Models through Self-Reflection
Large Language Models (LLMs) have been shown to exhibit various biases and stereotypes in their generated content. While extensive research has investigated bias in LLMs, prior work has predominantly focused on explicit bias, leaving the more nuanced implicit biases largely unexplored. This paper presents a systematic framework grounded in social psychology theories to investigate and compare explicit and implicit biases in LLMs. We propose a novel "self-reflection" based evaluation framework that operates in two phases: first measuring implicit bias through simulated psychological assessment methods, then evaluating explicit bias by prompting LLMs to analyze their own generated content. Through extensive experiments on state-of-the-art LLMs across multiple social dimensions, we demonstrate that LLMs exhibit a substantial inconsistency between explicit and implicit biases, where explicit biases manifest as mild stereotypes while implicit biases show strong stereotypes. Furthermore, we investigate the underlying factors contributing to this explicit-implicit bias inconsistency. Our experiments examine the effects of training data scale, model parameters, and alignment techniques. Results indicate that while explicit bias diminishes with increased training data and model size, implicit bias exhibits a contrasting upward trend. Notably, contemporary alignment methods (e.g., RLHF, DPO) effectively suppress explicit bias but show limited efficacy in mitigating implicit bias. These findings suggest that while scaling up models and alignment training can address explicit bias, the challenge of implicit bias requires novel approaches beyond current methodologies.
♻ ☆ NavRAG: Generating User Demand Instructions for Embodied Navigation through Retrieval-Augmented LLM
Vision-and-Language Navigation (VLN) is an essential skill for embodied agents, allowing them to navigate in 3D environments following natural language instructions. High-performance navigation models require a large amount of training data, the high cost of manually annotating data has seriously hindered this field. Therefore, some previous methods translate trajectory videos into step-by-step instructions for expanding data, but such instructions do not match well with users' communication styles that briefly describe destinations or state specific needs. Moreover, local navigation trajectories overlook global context and high-level task planning. To address these issues, we propose NavRAG, a retrieval-augmented generation (RAG) framework that generates user demand instructions for VLN. NavRAG leverages LLM to build a hierarchical scene description tree for 3D scene understanding from global layout to local details, then simulates various user roles with specific demands to retrieve from the scene tree, generating diverse instructions with LLM. We annotate over 2 million navigation instructions across 861 scenes and evaluate the data quality and navigation performance of trained models.
♻ ☆ How Diversely Can Language Models Solve Problems? Exploring the Algorithmic Diversity of Model-Generated Code
Language models (LMs) have exhibited impressive abilities in generating code from natural language requirements. In this work, we highlight the diversity of code generated by LMs as a critical criterion for evaluating their code generation capabilities. There is a lack of studies focused on assessing the diversity of generated code, which overlooks its importance in code LMs. Therefore, we propose a systematic approach to evaluate code diversity, introducing various metrics with inter-code similarity. Specifically, we introduce code clustering methods that leverages LMs' capabilities in code understanding and reasoning, resulting in a set of metrics that represent the number of algorithms in model-generated solutions. We extensively investigate the property of model-generated solutions by contrasting them with human-written ones and quantifying the impact of various factors on code diversity: model size, temperature, instruction tuning, and problem complexity. Our analysis demonstrates that model-generated solutions exhibit low algorithmic diversity, which was neglected by the research community. Moreover, we explore methods to increase code diversity by combining solutions from different models and increasing sampling temperatures. Our findings highlight that code diversity can be enhanced with the help of heterogeneous models and setting temperature beyond 1.0 that has not been fully explored due to the functional correctness degradation. To facilitate our research direction, we publicly share our code and datasets through open-source repositories.
♻ ☆ When Large Language Models Meet Evolutionary Algorithms: Potential Enhancements and Challenges
Pre-trained large language models (LLMs) exhibit powerful capabilities for generating natural text. Evolutionary algorithms (EAs) can discover diverse solutions to complex real-world problems. Motivated by the common collective and directionality of text generation and evolution, this paper first illustrates the conceptual parallels between LLMs and EAs at a micro level, which includes multiple one-to-one key characteristics: token representation and individual representation, position encoding and fitness shaping, position embedding and selection, Transformers block and reproduction, and model training and parameter adaptation. These parallels highlight potential opportunities for technical advancements in both LLMs and EAs. Subsequently, we analyze existing interdisciplinary research from a macro perspective to uncover critical challenges, with a particular focus on evolutionary fine-tuning and LLM-enhanced EAs. These analyses not only provide insights into the evolutionary mechanisms behind LLMs but also offer potential directions for enhancing the capabilities of artificial agents.
comment: The article has been accepted for publication in Research
♻ ☆ Zero-resource Hallucination Detection for Text Generation via Graph-based Contextual Knowledge Triples Modeling AAAI25
LLMs obtain remarkable performance but suffer from hallucinations. Most research on detecting hallucination focuses on the questions with short and concrete correct answers that are easy to check the faithfulness. Hallucination detections for text generation with open-ended answers are more challenging. Some researchers use external knowledge to detect hallucinations in generated texts, but external resources for specific scenarios are hard to access. Recent studies on detecting hallucinations in long text without external resources conduct consistency comparison among multiple sampled outputs. To handle long texts, researchers split long texts into multiple facts and individually compare the consistency of each pairs of facts. However, these methods (1) hardly achieve alignment among multiple facts; (2) overlook dependencies between multiple contextual facts. In this paper, we propose a graph-based context-aware (GCA) hallucination detection for text generations, which aligns knowledge facts and considers the dependencies between contextual knowledge triples in consistency comparison. Particularly, to align multiple facts, we conduct a triple-oriented response segmentation to extract multiple knowledge triples. To model dependencies among contextual knowledge triple (facts), we construct contextual triple into a graph and enhance triples' interactions via message passing and aggregating via RGCN. To avoid the omission of knowledge triples in long text, we conduct a LLM-based reverse verification via reconstructing the knowledge triples. Experiments show that our model enhances hallucination detection and excels all baselines.
comment: Accepted by AAAI25
♻ ☆ Energy-Based Diffusion Language Models for Text Generation
Despite remarkable progress in autoregressive language models, alternative generative paradigms beyond left-to-right generation are still being actively explored. Discrete diffusion models, with the capacity for parallel generation, have recently emerged as a promising alternative. Unfortunately, these models still underperform the autoregressive counterparts, with the performance gap increasing when reducing the number of sampling steps. Our analysis reveals that this degradation is a consequence of an imperfect approximation used by diffusion models. In this work, we propose Energy-based Diffusion Language Model (EDLM), an energy-based model operating at the full sequence level for each diffusion step, introduced to improve the underlying approximation used by diffusion models. More specifically, we introduce an EBM in a residual form, and show that its parameters can be obtained by leveraging a pretrained autoregressive model or by finetuning a bidirectional transformer via noise contrastive estimation. We also propose an efficient generation algorithm via parallel important sampling. Comprehensive experiments on language modeling benchmarks show that our model can consistently outperform state-of-the-art diffusion models by a significant margin, and approaches autoregressive models' perplexity. We further show that, without any generation performance drop, our framework offers a 1.3$\times$ sampling speedup over existing diffusion models. Reproduced code is available at https://github.com/MinkaiXu/Energy-Diffusion-LLM.
♻ ☆ Detection and Analysis of Offensive Online Content in Hausa Language
Hausa, a major Chadic language spoken by over 100 million people mostly in West Africa is considered a low-resource language from a computational linguistic perspective. This classification indicates a scarcity of linguistic resources and tools necessary for handling various natural language processing (NLP) tasks, including the detection of offensive content. To address this gap, we conducted two set of studies (1) a user study (n=101) to explore cyberbullying in Hausa and (2) an empirical study that led to the creation of the first dataset of offensive terms in the Hausa language. We developed detection systems trained on this dataset and compared their performance against relevant multilingual models, including Google Translate. Our detection system successfully identified over 70% of offensive, whereas baseline models frequently mistranslated such terms. We attribute this discrepancy to the nuanced nature of the Hausa language and the reliance of baseline models on direct or literal translation due to limited data to build purposive detection systems. These findings highlight the importance of incorporating cultural context and linguistic nuances when developing NLP models for low-resource languages such as Hausa. A post hoc analysis further revealed that offensive language is particularly prevalent in discussions related to religion and politics. To foster a safer online environment, we recommend involving diverse stakeholders with expertise in local contexts and demographics. Their insights will be crucial in developing more accurate detection systems and targeted moderation strategies that align with cultural sensitivities.
comment: 21 pages, 4 figures, 7 tables
Computer Vision and Pattern Recognition 140
☆ GoalFlow: Goal-Driven Flow Matching for Multimodal Trajectories Generation in End-to-End Autonomous Driving
We propose GoalFlow, an end-to-end autonomous driving method for generating high-quality multimodal trajectories. In autonomous driving scenarios, there is rarely a single suitable trajectory. Recent methods have increasingly focused on modeling multimodal trajectory distributions. However, they suffer from trajectory selection complexity and reduced trajectory quality due to high trajectory divergence and inconsistencies between guidance and scene information. To address these issues, we introduce GoalFlow, a novel method that effectively constrains the generative process to produce high-quality, multimodal trajectories. To resolve the trajectory divergence problem inherent in diffusion-based methods, GoalFlow constrains the generated trajectories by introducing a goal point. GoalFlow establishes a novel scoring mechanism that selects the most appropriate goal point from the candidate points based on scene information. Furthermore, GoalFlow employs an efficient generative method, Flow Matching, to generate multimodal trajectories, and incorporates a refined scoring mechanism to select the optimal trajectory from the candidates. Our experimental results, validated on the Navsim\cite{Dauner2024_navsim}, demonstrate that GoalFlow achieves state-of-the-art performance, delivering robust multimodal trajectories for autonomous driving. GoalFlow achieved PDMS of 90.3, significantly surpassing other methods. Compared with other diffusion-policy-based methods, our approach requires only a single denoising step to obtain excellent performance. The code is available at https://github.com/YvanYin/GoalFlow.
☆ Fairness-Aware Low-Rank Adaptation Under Demographic Privacy Constraints
Pre-trained foundation models can be adapted for specific tasks using Low-Rank Adaptation (LoRA). However, the fairness properties of these adapted classifiers remain underexplored. Existing fairness-aware fine-tuning methods rely on direct access to sensitive attributes or their predictors, but in practice, these sensitive attributes are often held under strict consumer privacy controls, and neither the attributes nor their predictors are available to model developers, hampering the development of fair models. To address this issue, we introduce a set of LoRA-based fine-tuning methods that can be trained in a distributed fashion, where model developers and fairness auditors collaborate without sharing sensitive attributes or predictors. In this paper, we evaluate three such methods - sensitive unlearning, adversarial training, and orthogonality loss - against a fairness-unaware baseline, using experiments on the CelebA and UTK-Face datasets with an ImageNet pre-trained ViT-Base model. We find that orthogonality loss consistently reduces bias while maintaining or improving utility, whereas adversarial training improves False Positive Rate Parity and Demographic Parity in some cases, and sensitive unlearning provides no clear benefit. In tasks where significant biases are present, distributed fairness-aware fine-tuning methods can effectively eliminate bias without compromising consumer privacy and, in most cases, improve model utility.
☆ Task-oriented Uncertainty Collaborative Learning for Label-Efficient Brain Tumor Segmentation
Multi-contrast magnetic resonance imaging (MRI) plays a vital role in brain tumor segmentation and diagnosis by leveraging complementary information from different contrasts. Each contrast highlights specific tumor characteristics, enabling a comprehensive understanding of tumor morphology, edema, and pathological heterogeneity. However, existing methods still face the challenges of multi-level specificity perception across different contrasts, especially with limited annotations. These challenges include data heterogeneity, granularity differences, and interference from redundant information. To address these limitations, we propose a Task-oriented Uncertainty Collaborative Learning (TUCL) framework for multi-contrast MRI segmentation. TUCL introduces a task-oriented prompt attention (TPA) module with intra-prompt and cross-prompt attention mechanisms to dynamically model feature interactions across contrasts and tasks. Additionally, a cyclic process is designed to map the predictions back to the prompt to ensure that the prompts are effectively utilized. In the decoding stage, the TUCL framework proposes a dual-path uncertainty refinement (DUR) strategy which ensures robust segmentation by refining predictions iteratively. Extensive experimental results on limited labeled data demonstrate that TUCL significantly improves segmentation accuracy (88.2\% in Dice and 10.853 mm in HD95). It shows that TUCL has the potential to extract multi-contrast information and reduce the reliance on extensive annotations. The code is available at: https://github.com/Zhenxuan-Zhang/TUCL_BrainSeg.
☆ AIM-Fair: Advancing Algorithmic Fairness via Selectively Fine-Tuning Biased Models with Contextual Synthetic Data CVPR 2025
Recent advances in generative models have sparked research on improving model fairness with AI-generated data. However, existing methods often face limitations in the diversity and quality of synthetic data, leading to compromised fairness and overall model accuracy. Moreover, many approaches rely on the availability of demographic group labels, which are often costly to annotate. This paper proposes AIM-Fair, aiming to overcome these limitations and harness the potential of cutting-edge generative models in promoting algorithmic fairness. We investigate a fine-tuning paradigm starting from a biased model initially trained on real-world data without demographic annotations. This model is then fine-tuned using unbiased synthetic data generated by a state-of-the-art diffusion model to improve its fairness. Two key challenges are identified in this fine-tuning paradigm, 1) the low quality of synthetic data, which can still happen even with advanced generative models, and 2) the domain and bias gap between real and synthetic data. To address the limitation of synthetic data quality, we propose Contextual Synthetic Data Generation (CSDG) to generate data using a text-to-image diffusion model (T2I) with prompts generated by a context-aware LLM, ensuring both data diversity and control of bias in synthetic data. To resolve domain and bias shifts, we introduce a novel selective fine-tuning scheme in which only model parameters more sensitive to bias and less sensitive to domain shift are updated. Experiments on CelebA and UTKFace datasets show that our AIM-Fair improves model fairness while maintaining utility, outperforming both fully and partially fine-tuned approaches to model fairness.
comment: Accepted at CVPR 2025. Github: https://github.com/zengqunzhao/AIM-Fair. Project page: https://zengqunzhao.github.io/AIMFair
☆ NoT: Federated Unlearning via Weight Negation
Federated unlearning (FU) aims to remove a participant's data contributions from a trained federated learning (FL) model, ensuring privacy and regulatory compliance. Traditional FU methods often depend on auxiliary storage on either the client or server side or require direct access to the data targeted for removal-a dependency that may not be feasible if the data is no longer available. To overcome these limitations, we propose NoT, a novel and efficient FU algorithm based on weight negation (multiplying by -1), which circumvents the need for additional storage and access to the target data. We argue that effective and efficient unlearning can be achieved by perturbing model parameters away from the set of optimal parameters, yet being well-positioned for quick re-optimization. This technique, though seemingly contradictory, is theoretically grounded: we prove that the weight negation perturbation effectively disrupts inter-layer co-adaptation, inducing unlearning while preserving an approximate optimality property, thereby enabling rapid recovery. Experimental results across three datasets and three model architectures demonstrate that NoT significantly outperforms existing baselines in unlearning efficacy as well as in communication and computational efficiency.
comment: The 42nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville TN, US. 2025
☆ BEHAVIOR Robot Suite: Streamlining Real-World Whole-Body Manipulation for Everyday Household Activities
Real-world household tasks present significant challenges for mobile manipulation robots. An analysis of existing robotics benchmarks reveals that successful task performance hinges on three key whole-body control capabilities: bimanual coordination, stable and precise navigation, and extensive end-effector reachability. Achieving these capabilities requires careful hardware design, but the resulting system complexity further complicates visuomotor policy learning. To address these challenges, we introduce the BEHAVIOR Robot Suite (BRS), a comprehensive framework for whole-body manipulation in diverse household tasks. Built on a bimanual, wheeled robot with a 4-DoF torso, BRS integrates a cost-effective whole-body teleoperation interface for data collection and a novel algorithm for learning whole-body visuomotor policies. We evaluate BRS on five challenging household tasks that not only emphasize the three core capabilities but also introduce additional complexities, such as long-range navigation, interaction with articulated and deformable objects, and manipulation in confined spaces. We believe that BRS's integrated robotic embodiment, data collection interface, and learning framework mark a significant step toward enabling real-world whole-body manipulation for everyday household tasks. BRS is open-sourced at https://behavior-robot-suite.github.io/
comment: Project website: https://behavior-robot-suite.github.io/
☆ VideoPainter: Any-length Video Inpainting and Editing with Plug-and-Play Context Control
Video inpainting, which aims to restore corrupted video content, has experienced substantial progress. Despite these advances, existing methods, whether propagating unmasked region pixels through optical flow and receptive field priors, or extending image-inpainting models temporally, face challenges in generating fully masked objects or balancing the competing objectives of background context preservation and foreground generation in one model, respectively. To address these limitations, we propose a novel dual-stream paradigm VideoPainter that incorporates an efficient context encoder (comprising only 6% of the backbone parameters) to process masked videos and inject backbone-aware background contextual cues to any pre-trained video DiT, producing semantically consistent content in a plug-and-play manner. This architectural separation significantly reduces the model's learning complexity while enabling nuanced integration of crucial background context. We also introduce a novel target region ID resampling technique that enables any-length video inpainting, greatly enhancing our practical applicability. Additionally, we establish a scalable dataset pipeline leveraging current vision understanding models, contributing VPData and VPBench to facilitate segmentation-based inpainting training and assessment, the largest video inpainting dataset and benchmark to date with over 390K diverse clips. Using inpainting as a pipeline basis, we also explore downstream applications including video editing and video editing pair data generation, demonstrating competitive performance and significant practical potential. Extensive experiments demonstrate VideoPainter's superior performance in both any-length video inpainting and editing, across eight key metrics, including video quality, mask region preservation, and textual coherence.
comment: Project page available at https://yxbian23.github.io/project/video-painter
☆ TrajectoryCrafter: Redirecting Camera Trajectory for Monocular Videos via Diffusion Models
We present TrajectoryCrafter, a novel approach to redirect camera trajectories for monocular videos. By disentangling deterministic view transformations from stochastic content generation, our method achieves precise control over user-specified camera trajectories. We propose a novel dual-stream conditional video diffusion model that concurrently integrates point cloud renders and source videos as conditions, ensuring accurate view transformations and coherent 4D content generation. Instead of leveraging scarce multi-view videos, we curate a hybrid training dataset combining web-scale monocular videos with static multi-view datasets, by our innovative double-reprojection strategy, significantly fostering robust generalization across diverse scenes. Extensive evaluations on multi-view and large-scale monocular videos demonstrate the superior performance of our method.
comment: Project webpage: https://trajectorycrafter.github.io/
☆ Joint 3D Point Cloud Segmentation using Real-Sim Loop: From Panels to Trees and Branches ICRA 2025
Modern orchards are planted in structured rows with distinct panel divisions to improve management. Accurate and efficient joint segmentation of point cloud from Panel to Tree and Branch (P2TB) is essential for robotic operations. However, most current segmentation methods focus on single instance segmentation and depend on a sequence of deep networks to perform joint tasks. This strategy hinders the use of hierarchical information embedded in the data, leading to both error accumulation and increased costs for annotation and computation, which limits its scalability for real-world applications. In this study, we proposed a novel approach that incorporated a Real2Sim L-TreeGen for training data generation and a joint model (J-P2TB) designed for the P2TB task. The J-P2TB model, trained on the generated simulation dataset, was used for joint segmentation of real-world panel point clouds via zero-shot learning. Compared to representative methods, our model outperformed them in most segmentation metrics while using 40% fewer learnable parameters. This Sim2Real result highlighted the efficacy of L-TreeGen in model training and the performance of J-P2TB for joint segmentation, demonstrating its strong accuracy, efficiency, and generalizability for real-world applications. These improvements would not only greatly benefit the development of robots for automated orchard operations but also advance digital twin technology.
comment: Accepted by ICRA 2025
☆ FMT:A Multimodal Pneumonia Detection Model Based on Stacking MOE Framework
Artificial intelligence has shown the potential to improve diagnostic accuracy through medical image analysis for pneumonia diagnosis. However, traditional multimodal approaches often fail to address real-world challenges such as incomplete data and modality loss. In this study, a Flexible Multimodal Transformer (FMT) was proposed, which uses ResNet-50 and BERT for joint representation learning, followed by a dynamic masked attention strategy that simulates clinical modality loss to improve robustness; finally, a sequential mixture of experts (MOE) architecture was used to achieve multi-level decision refinement. After evaluation on a small multimodal pneumonia dataset, FMT achieved state-of-the-art performance with 94% accuracy, 95% recall, and 93% F1 score, outperforming single-modal baselines (ResNet: 89%; BERT: 79%) and the medical benchmark CheXMed (90%), providing a scalable solution for multimodal diagnosis of pneumonia in resource-constrained medical settings.
☆ Conformal Prediction for Image Segmentation Using Morphological Prediction Sets
Image segmentation is a challenging task influenced by multiple sources of uncertainty, such as the data labeling process or the sampling of training data. In this paper we focus on binary segmentation and address these challenges using conformal prediction, a family of model- and data-agnostic methods for uncertainty quantification that provide finite-sample theoretical guarantees and applicable to any pretrained predictor. Our approach involves computing nonconformity scores, a type of prediction residual, on held-out calibration data not used during training. We use dilation, one of the fundamental operations in mathematical morphology, to construct a margin added to the borders of predicted segmentation masks. At inference, the predicted set formed by the mask and its margin contains the ground-truth mask with high probability, at a confidence level specified by the user. The size of the margin serves as an indicator of predictive uncertainty for a given model and dataset. We work in a regime of minimal information as we do not require any feedback from the predictor: only the predicted masks are needed for computing the prediction sets. Hence, our method is applicable to any segmentation model, including those based on deep learning; we evaluate our approach on several medical imaging applications.
☆ CACTUS: An Open Dataset and Framework for Automated Cardiac Assessment and Classification of Ultrasound Images Using Deep Transfer Learning
Cardiac ultrasound (US) scanning is a commonly used techniques in cardiology to diagnose the health of the heart and its proper functioning. Therefore, it is necessary to consider ways to automate these tasks and assist medical professionals in classifying and assessing cardiac US images. Machine learning (ML) techniques are regarded as a prominent solution due to their success in numerous applications aimed at enhancing the medical field, including addressing the shortage of echography technicians. However, the limited availability of medical data presents a significant barrier to applying ML in cardiology, particularly regarding US images of the heart. This paper addresses this challenge by introducing the first open graded dataset for Cardiac Assessment and ClassificaTion of UltraSound (CACTUS), which is available online. This dataset contains images obtained from scanning a CAE Blue Phantom and representing various heart views and different quality levels, exceeding the conventional cardiac views typically found in the literature. Additionally, the paper introduces a Deep Learning (DL) framework consisting of two main components. The first component classifies cardiac US images based on the heart view using a Convolutional Neural Network (CNN). The second component uses Transfer Learning (TL) to fine-tune the knowledge from the first component and create a model for grading and assessing cardiac images. The framework demonstrates high performance in both classification and grading, achieving up to 99.43% accuracy and as low as 0.3067 error, respectively. To showcase its robustness, the framework is further fine-tuned using new images representing additional cardiac views and compared to several other state-of-the-art architectures. The framework's outcomes and performance in handling real-time scans were also assessed using a questionnaire answered by cardiac experts.
☆ D2GV: Deformable 2D Gaussian Splatting for Video Representation in 400FPS
Implicit Neural Representations (INRs) have emerged as a powerful approach for video representation, offering versatility across tasks such as compression and inpainting. However, their implicit formulation limits both interpretability and efficacy, undermining their practicality as a comprehensive solution. We propose a novel video representation based on deformable 2D Gaussian splatting, dubbed D2GV, which aims to achieve three key objectives: 1) improved efficiency while delivering superior quality; 2) enhanced scalability and interpretability; and 3) increased friendliness for downstream tasks. Specifically, we initially divide the video sequence into fixed-length Groups of Pictures (GoP) to allow parallel training and linear scalability with video length. For each GoP, D2GV represents video frames by applying differentiable rasterization to 2D Gaussians, which are deformed from a canonical space into their corresponding timestamps. Notably, leveraging efficient CUDA-based rasterization, D2GV converges fast and decodes at speeds exceeding 400 FPS, while delivering quality that matches or surpasses state-of-the-art INRs. Moreover, we incorporate a learnable pruning and quantization strategy to streamline D2GV into a more compact representation. We demonstrate D2GV's versatility in tasks including video interpolation, inpainting and denoising, underscoring its potential as a promising solution for video representation. Code is available at: \href{https://github.com/Evan-sudo/D2GV}{https://github.com/Evan-sudo/D2GV}.
☆ Anti-Diffusion: Preventing Abuse of Modifications of Diffusion-Based Models
Although diffusion-based techniques have shown remarkable success in image generation and editing tasks, their abuse can lead to severe negative social impacts. Recently, some works have been proposed to provide defense against the abuse of diffusion-based methods. However, their protection may be limited in specific scenarios by manually defined prompts or the stable diffusion (SD) version. Furthermore, these methods solely focus on tuning methods, overlooking editing methods that could also pose a significant threat. In this work, we propose Anti-Diffusion, a privacy protection system designed for general diffusion-based methods, applicable to both tuning and editing techniques. To mitigate the limitations of manually defined prompts on defense performance, we introduce the prompt tuning (PT) strategy that enables precise expression of original images. To provide defense against both tuning and editing methods, we propose the semantic disturbance loss (SDL) to disrupt the semantic information of protected images. Given the limited research on the defense against editing methods, we develop a dataset named Defense-Edit to assess the defense performance of various methods. Experiments demonstrate that our Anti-Diffusion achieves superior defense performance across a wide range of diffusion-based techniques in different scenarios.
QArtSR: Quantization via Reverse-Module and Timestep-Retraining in One-Step Diffusion based Image Super-Resolution
One-step diffusion-based image super-resolution (OSDSR) models are showing increasingly superior performance nowadays. However, although their denoising steps are reduced to one and they can be quantized to 8-bit to reduce the costs further, there is still significant potential for OSDSR to quantize to lower bits. To explore more possibilities of quantized OSDSR, we propose an efficient method, Quantization via reverse-module and timestep-retraining for OSDSR, named QArtSR. Firstly, we investigate the influence of timestep value on the performance of quantized models. Then, we propose Timestep Retraining Quantization (TRQ) and Reversed Per-module Quantization (RPQ) strategies to calibrate the quantized model. Meanwhile, we adopt the module and image losses to update all quantized modules. We only update the parameters in quantization finetuning components, excluding the original weights. To ensure that all modules are fully finetuned, we add extended end-to-end training after per-module stage. Our 4-bit and 2-bit quantization experimental results indicate that QArtSR obtains superior effects against the recent leading comparison methods. The performance of 4-bit QArtSR is close to the full-precision one. Our code will be released at https://github.com/libozhu03/QArtSR.
☆ Novel Object 6D Pose Estimation with a Single Reference View
Existing novel object 6D pose estimation methods typically rely on CAD models or dense reference views, which are both difficult to acquire. Using only a single reference view is more scalable, but challenging due to large pose discrepancies and limited geometric and spatial information. To address these issues, we propose a Single-Reference-based novel object 6D (SinRef-6D) pose estimation method. Our key idea is to iteratively establish point-wise alignment in the camera coordinate system based on state space models (SSMs). Specifically, iterative camera-space point-wise alignment can effectively handle large pose discrepancies, while our proposed RGB and Points SSMs can capture long-range dependencies and spatial information from a single view, offering linear complexity and superior spatial modeling capability. Once pre-trained on synthetic data, SinRef-6D can estimate the 6D pose of a novel object using only a single reference view, without requiring retraining or a CAD model. Extensive experiments on six popular datasets and real-world robotic scenes demonstrate that we achieve on-par performance with CAD-based and dense reference view-based methods, despite operating in the more challenging single reference setting. Code will be released at https://github.com/CNJianLiu/SinRef-6D.
comment: 17 pages, 12 figures (including supplementary material)
☆ TomatoScanner: phenotyping tomato fruit based on only RGB image
In tomato greenhouse, phenotypic measurement is meaningful for researchers and farmers to monitor crop growth, thereby precisely control environmental conditions in time, leading to better quality and higher yield. Traditional phenotyping mainly relies on manual measurement, which is accurate but inefficient, more importantly, endangering the health and safety of people. Several studies have explored computer vision-based methods to replace manual phenotyping. However, the 2D-based need extra calibration, or cause destruction to fruit, or can only measure limited and meaningless traits. The 3D-based need extra depth camera, which is expensive and unacceptable for most farmers. In this paper, we propose a non-contact tomato fruit phenotyping method, titled TomatoScanner, where RGB image is all you need for input. First, pixel feature is extracted by instance segmentation of our proposed EdgeYOLO with preprocessing of individual separation and pose correction. Second, depth feature is extracted by depth estimation of Depth Pro. Third, pixel and depth feature are fused to output phenotype results in reality. We establish self-built Tomato Phenotype Dataset to test TomatoScanner, which achieves excellent phenotyping on width, height, vertical area and volume, with median relative error of 5.63%, 7.03%, -0.64% and 37.06%, respectively. We propose and add three innovative modules - EdgeAttention, EdgeLoss and EdgeBoost - into EdgeYOLO, to enhance the segmentation accuracy on edge portion. Precision and mean Edge Error greatly improve from 0.943 and 5.641% to 0.986 and 2.963%, respectively. Meanwhile, EdgeYOLO keeps lightweight and efficient, with 48.7 M weights size and 76.34 FPS. Codes and datasets: https://github.com/AlexTraveling/TomatoScanner.
comment: 12 pages, 37 figures. Codes and datasets are open-sourced in https://github.com/AlexTraveling/TomatoScanner
☆ Stereo Any Video: Temporally Consistent Stereo Matching
This paper introduces Stereo Any Video, a powerful framework for video stereo matching. It can estimate spatially accurate and temporally consistent disparities without relying on auxiliary information such as camera poses or optical flow. The strong capability is driven by rich priors from monocular video depth models, which are integrated with convolutional features to produce stable representations. To further enhance performance, key architectural innovations are introduced: all-to-all-pairs correlation, which constructs smooth and robust matching cost volumes, and temporal convex upsampling, which improves temporal coherence. These components collectively ensure robustness, accuracy, and temporal consistency, setting a new standard in video stereo matching. Extensive experiments demonstrate that our method achieves state-of-the-art performance across multiple datasets both qualitatively and quantitatively in zero-shot settings, as well as strong generalization to real-world indoor and outdoor scenarios.
☆ Pi-GPS: Enhancing Geometry Problem Solving by Unleashing the Power of Diagrammatic Information
Geometry problem solving has garnered increasing attention due to its potential applications in intelligent education field. Inspired by the observation that text often introduces ambiguities that diagrams can clarify, this paper presents Pi-GPS, a novel framework that unleashes the power of diagrammatic information to resolve textual ambiguities, an aspect largely overlooked in prior research. Specifically, we design a micro module comprising a rectifier and verifier: the rectifier employs MLLMs to disambiguate text based on the diagrammatic context, while the verifier ensures the rectified output adherence to geometric rules, mitigating model hallucinations. Additionally, we explore the impact of LLMs in theorem predictor based on the disambiguated formal language. Empirical results demonstrate that Pi-GPS surpasses state-of-the-art models, achieving a nearly 10\% improvement on Geometry3K over prior neural-symbolic approaches. We hope this work highlights the significance of resolving textual ambiguity in multimodal mathematical reasoning, a crucial factor limiting performance.
☆ Disconnect to Connect: A Data Augmentation Method for Improving Topology Accuracy in Image Segmentation
Accurate segmentation of thin, tubular structures (e.g., blood vessels) is challenging for deep neural networks. These networks classify individual pixels, and even minor misclassifications can break the thin connections within these structures. Existing methods for improving topology accuracy, such as topology loss functions, rely on very precise, topologically-accurate training labels, which are difficult to obtain. This is because annotating images, especially 3D images, is extremely laborious and time-consuming. Low image resolution and contrast further complicates the annotation by causing tubular structures to appear disconnected. We present CoLeTra, a data augmentation strategy that integrates to the models the prior knowledge that structures that appear broken are actually connected. This is achieved by creating images with the appearance of disconnected structures while maintaining the original labels. Our extensive experiments, involving different architectures, loss functions, and datasets, demonstrate that CoLeTra leads to segmentations topologically more accurate while often improving the Dice coefficient and Hausdorff distance. CoLeTra's hyper-parameters are intuitive to tune, and our sensitivity analysis shows that CoLeTra is robust to changes in these hyper-parameters. We also release a dataset specifically suited for image segmentation methods with a focus on topology accuracy. CoLetra's code can be found at https://github.com/jmlipman/CoLeTra.
☆ S4M: Segment Anything with 4 Extreme Points
The Segment Anything Model (SAM) has revolutionized open-set interactive image segmentation, inspiring numerous adapters for the medical domain. However, SAM primarily relies on sparse prompts such as point or bounding box, which may be suboptimal for fine-grained instance segmentation, particularly in endoscopic imagery, where precise localization is critical and existing prompts struggle to capture object boundaries effectively. To address this, we introduce S4M (Segment Anything with 4 Extreme Points), which augments SAM by leveraging extreme points -- the top-, bottom-, left-, and right-most points of an instance -- prompts. These points are intuitive to identify and provide a faster, structured alternative to box prompts. However, a na\"ive use of extreme points degrades performance, due to SAM's inability to interpret their semantic roles. To resolve this, we introduce dedicated learnable embeddings, enabling the model to distinguish extreme points from generic free-form points and better reason about their spatial relationships. We further propose an auxiliary training task through the Canvas module, which operates solely on prompts -- without vision input -- to predict a coarse instance mask. This encourages the model to internalize the relationship between extreme points and mask distributions, leading to more robust segmentation. S4M outperforms other SAM-based approaches on three endoscopic surgical datasets, demonstrating its effectiveness in complex scenarios. Finally, we validate our approach through a human annotation study on surgical endoscopic videos, confirming that extreme points are faster to acquire than bounding boxes.
☆ State-of-the-Art Stroke Lesion Segmentation at 1/1000th of Parameters
Efficient and accurate whole-brain lesion segmentation remains a challenge in medical image analysis. In this work, we revisit MeshNet, a parameter-efficient segmentation model, and introduce a novel multi-scale dilation pattern with an encoder-decoder structure. This innovation enables capturing broad contextual information and fine-grained details without traditional downsampling, upsampling, or skip-connections. Unlike previous approaches processing subvolumes or slices, we operate directly on whole-brain $256^3$ MRI volumes. Evaluations on the Aphasia Recovery Cohort (ARC) dataset demonstrate that MeshNet achieves superior or comparable DICE scores to state-of-the-art architectures such as MedNeXt and U-MAMBA at 1/1000th of parameters. Our results validate MeshNet's strong balance of efficiency and performance, making it particularly suitable for resource-limited environments such as web-based applications and opening new possibilities for the widespread deployment of advanced medical image analysis tools.
comment: International Symposium on Biomedical Imaging, April 14-17, 2025
☆ Post-Hoc Concept Disentanglement: From Correlated to Isolated Concept Representations
Concept Activation Vectors (CAVs) are widely used to model human-understandable concepts as directions within the latent space of neural networks. They are trained by identifying directions from the activations of concept samples to those of non-concept samples. However, this method often produces similar, non-orthogonal directions for correlated concepts, such as "beard" and "necktie" within the CelebA dataset, which frequently co-occur in images of men. This entanglement complicates the interpretation of concepts in isolation and can lead to undesired effects in CAV applications, such as activation steering. To address this issue, we introduce a post-hoc concept disentanglement method that employs a non-orthogonality loss, facilitating the identification of orthogonal concept directions while preserving directional correctness. We evaluate our approach with real-world and controlled correlated concepts in CelebA and a synthetic FunnyBirds dataset with VGG16 and ResNet18 architectures. We further demonstrate the superiority of orthogonalized concept representations in activation steering tasks, allowing (1) the insertion of isolated concepts into input images through generative models and (2) the removal of concepts for effective shortcut suppression with reduced impact on correlated concepts in comparison to baseline CAVs.
☆ Removing Geometric Bias in One-Class Anomaly Detection with Adaptive Feature Perturbation WACV 2025
One-class anomaly detection aims to detect objects that do not belong to a predefined normal class. In practice training data lack those anomalous samples; hence state-of-the-art methods are trained to discriminate between normal and synthetically-generated pseudo-anomalous data. Most methods use data augmentation techniques on normal images to simulate anomalies. However the best-performing ones implicitly leverage a geometric bias present in the benchmarking datasets. This limits their usability in more general conditions. Others are relying on basic noising schemes that may be suboptimal in capturing the underlying structure of normal data. In addition most still favour the image domain to generate pseudo-anomalies training models end-to-end from only the normal class and overlooking richer representations of the information. To overcome these limitations we consider frozen yet rich feature spaces given by pretrained models and create pseudo-anomalous features with a novel adaptive linear feature perturbation technique. It adapts the noise distribution to each sample applies decaying linear perturbations to feature vectors and further guides the classification process using a contrastive learning objective. Experimental evaluation conducted on both standard and geometric bias-free datasets demonstrates the superiority of our approach with respect to comparable baselines. The codebase is accessible via our public repository.
comment: Published in WACV 2025
☆ FastMap: Fast Queries Initialization Based Vectorized HD Map Reconstruction Framework
Reconstruction of high-definition maps is a crucial task in perceiving the autonomous driving environment, as its accuracy directly impacts the reliability of prediction and planning capabilities in downstream modules. Current vectorized map reconstruction methods based on the DETR framework encounter limitations due to the redundancy in the decoder structure, necessitating the stacking of six decoder layers to maintain performance, which significantly hampers computational efficiency. To tackle this issue, we introduce FastMap, an innovative framework designed to reduce decoder redundancy in existing approaches. FastMap optimizes the decoder architecture by employing a single-layer, two-stage transformer that achieves multilevel representation capabilities. Our framework eliminates the conventional practice of randomly initializing queries and instead incorporates a heatmap-guided query generation module during the decoding phase, which effectively maps image features into structured query vectors using learnable positional encoding. Additionally, we propose a geometry-constrained point-to-line loss mechanism for FastMap, which adeptly addresses the challenge of distinguishing highly homogeneous features that often arise in traditional point-to-point loss computations. Extensive experiments demonstrate that FastMap achieves state-of-the-art performance in both nuScenes and Argoverse2 datasets, with its decoder operating 3.2 faster than the baseline. Code and more demos are available at https://github.com/hht1996ok/FastMap.
☆ DecoupledGaussian: Object-Scene Decoupling for Physics-Based Interaction CVPR2025
We present DecoupledGaussian, a novel system that decouples static objects from their contacted surfaces captured in-the-wild videos, a key prerequisite for realistic Newtonian-based physical simulations. Unlike prior methods focused on synthetic data or elastic jittering along the contact surface, which prevent objects from fully detaching or moving independently, DecoupledGaussian allows for significant positional changes without being constrained by the initial contacted surface. Recognizing the limitations of current 2D inpainting tools for restoring 3D locations, our approach proposes joint Poisson fields to repair and expand the Gaussians of both objects and contacted scenes after separation. This is complemented by a multi-carve strategy to refine the object's geometry. Our system enables realistic simulations of decoupling motions, collisions, and fractures driven by user-specified impulses, supporting complex interactions within and across multiple scenes. We validate DecoupledGaussian through a comprehensive user study and quantitative benchmarks. This system enhances digital interaction with objects and scenes in real-world environments, benefiting industries such as VR, robotics, and autonomous driving. Our project page is at: https://wangmiaowei.github.io/DecoupledGaussian.github.io/.
comment: CVPR2025 Accepted
☆ Automatic Teaching Platform on Vision Language Retrieval Augmented Generation
Automating teaching presents unique challenges, as replicating human interaction and adaptability is complex. Automated systems cannot often provide nuanced, real-time feedback that aligns with students' individual learning paces or comprehension levels, which can hinder effective support for diverse needs. This is especially challenging in fields where abstract concepts require adaptive explanations. In this paper, we propose a vision language retrieval augmented generation (named VL-RAG) system that has the potential to bridge this gap by delivering contextually relevant, visually enriched responses that can enhance comprehension. By leveraging a database of tailored answers and images, the VL-RAG system can dynamically retrieve information aligned with specific questions, creating a more interactive and engaging experience that fosters deeper understanding and active student participation. It allows students to explore concepts visually and verbally, promoting deeper understanding and reducing the need for constant human oversight while maintaining flexibility to expand across different subjects and course material.
☆ Towards Locally Explaining Prediction Behavior via Gradual Interventions and Measuring Property Gradients
Deep learning models achieve high predictive performance but lack intrinsic interpretability, hindering our understanding of the learned prediction behavior. Existing local explainability methods focus on associations, neglecting the causal drivers of model predictions. Other approaches adopt a causal perspective but primarily provide more general global explanations. However, for specific inputs, it's unclear whether globally identified factors apply locally. To address this limitation, we introduce a novel framework for local interventional explanations by leveraging recent advances in image-to-image editing models. Our approach performs gradual interventions on semantic properties to quantify the corresponding impact on a model's predictions using a novel score, the expected property gradient magnitude. We demonstrate the effectiveness of our approach through an extensive empirical evaluation on a wide range of architectures and tasks. First, we validate it in a synthetic scenario and demonstrate its ability to locally identify biases. Afterward, we apply our approach to analyze network training dynamics, investigate medical skin lesion classifiers, and study a pre-trained CLIP model with real-life interventional data. Our results highlight the potential of interventional explanations on the property level to reveal new insights into the behavior of deep models.
comment: 44 pages, 39 figures, 14 tables
☆ Semantic Shift Estimation via Dual-Projection and Classifier Reconstruction for Exemplar-Free Class-Incremental Learning
Exemplar-Free Class-Incremental Learning (EFCIL) aims to sequentially learn from distinct categories without retaining exemplars but easily suffers from catastrophic forgetting of learned knowledge. While existing EFCIL methods leverage knowledge distillation to alleviate forgetting, they still face two critical challenges: semantic shift and decision bias. Specifically, the embeddings of old tasks shift in the embedding space after learning new tasks, and the classifier becomes biased towards new tasks due to training solely with new data, thereby hindering the balance between old and new knowledge. To address these issues, we propose the Dual-Projection Shift Estimation and Classifier Reconstruction (DPCR) approach for EFCIL. DPCR effectively estimates semantic shift through a dual-projection, which combines a learnable transformation with a row-space projection to capture both task-wise and category-wise shifts. Furthermore, to mitigate decision bias, DPCR employs ridge regression to reformulate classifier training as a reconstruction process. This reconstruction exploits previous information encoded in covariance and prototype of each class after calibration with estimated shift, thereby reducing decision bias. Extensive experiments demonstrate that, across various datasets, DPCR effectively balances old and new tasks, outperforming state-of-the-art EFCIL methods.
comment: 14 pages, 7 figures
☆ Self-Modeling Robots by Photographing
Self-modeling enables robots to build task-agnostic models of their morphology and kinematics based on data that can be automatically collected, with minimal human intervention and prior information, thereby enhancing machine intelligence. Recent research has highlighted the potential of data-driven technology in modeling the morphology and kinematics of robots. However, existing self-modeling methods suffer from either low modeling quality or excessive data acquisition costs. Beyond morphology and kinematics, texture is also a crucial component of robots, which is challenging to model and remains unexplored. In this work, a high-quality, texture-aware, and link-level method is proposed for robot self-modeling. We utilize three-dimensional (3D) Gaussians to represent the static morphology and texture of robots, and cluster the 3D Gaussians to construct neural ellipsoid bones, whose deformations are controlled by the transformation matrices generated by a kinematic neural network. The 3D Gaussians and kinematic neural network are trained using data pairs composed of joint angles, camera parameters and multi-view images without depth information. By feeding the kinematic neural network with joint angles, we can utilize the well-trained model to describe the corresponding morphology, kinematics and texture of robots at the link level, and render robot images from different perspectives with the aid of 3D Gaussian splatting. Furthermore, we demonstrate that the established model can be exploited to perform downstream tasks such as motion planning and inverse kinematics.
☆ R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcing Learning
In this work, we present the first application of Reinforcement Learning with Verifiable Reward (RLVR) to an Omni-multimodal large language model in the context of emotion recognition, a task where both visual and audio modalities play crucial roles. We leverage RLVR to optimize the Omni model, significantly enhancing its performance in three key aspects: reasoning capability, emotion recognition accuracy, and generalization ability. The introduction of RLVR not only improves the model's overall performance on in-distribution data but also demonstrates superior robustness when evaluated on out-of-distribution datasets. More importantly, the improved reasoning capability enables clear analysis of the contributions of different modalities, particularly visual and audio information, in the emotion recognition process. This provides valuable insights into the optimization of multimodal large language models.
☆ Multi-Grained Feature Pruning for Video-Based Human Pose Estimation
Human pose estimation, with its broad applications in action recognition and motion capture, has experienced significant advancements. However, current Transformer-based methods for video pose estimation often face challenges in managing redundant temporal information and achieving fine-grained perception because they only focus on processing low-resolution features. To address these challenges, we propose a novel multi-scale resolution framework that encodes spatio-temporal representations at varying granularities and executes fine-grained perception compensation. Furthermore, we employ a density peaks clustering method to dynamically identify and prioritize tokens that offer important semantic information. This strategy effectively prunes redundant feature tokens, especially those arising from multi-frame features, thereby optimizing computational efficiency without sacrificing semantic richness. Empirically, it sets new benchmarks for both performance and efficiency on three large-scale datasets. Our method achieves a 93.8% improvement in inference speed compared to the baseline, while also enhancing pose estimation accuracy, reaching 87.4 mAP on the PoseTrack2017 dataset.
☆ Pretext Task Adversarial Learning for Unpaired Low-field to Ultra High-field MRI Synthesis
Given the scarcity and cost of high-field MRI, the synthesis of high-field MRI from low-field MRI holds significant potential when there is limited data for training downstream tasks (e.g. segmentation). Low-field MRI often suffers from a reduced signal-to-noise ratio (SNR) and spatial resolution compared to high-field MRI. However, synthesizing high-field MRI data presents challenges. These involve aligning image features across domains while preserving anatomical accuracy and enhancing fine details. To address these challenges, we propose a Pretext Task Adversarial (PTA) learning framework for high-field MRI synthesis from low-field MRI data. The framework comprises three processes: (1) The slice-wise gap perception (SGP) network aligns the slice inconsistencies of low-field and high-field datasets based on contrastive learning. (2) The local structure correction (LSC) network extracts local structures by restoring the locally rotated and masked images. (3) The pretext task-guided adversarial training process introduces additional supervision and incorporates a discriminator to improve image realism. Extensive experiments on low-field to ultra high-field task demonstrate the effectiveness of our method, achieving state-of-the-art performance (16.892 in FID, 1.933 in IS, and 0.324 in MS-SSIM). This enables the generation of high-quality high-field-like MRI data from low-field MRI data to augment training datasets for downstream tasks. The code is available at: https://github.com/Zhenxuan-Zhang/PTA4Unpaired_HF_MRI_SYN.
☆ New multimodal similarity measure for image registration via modeling local functional dependence with linear combination of learned basis functions
The deformable registration of images of different modalities, essential in many medical imaging applications, remains challenging. The main challenge is developing a robust measure for image overlap despite the compared images capturing different aspects of the underlying tissue. Here, we explore similarity metrics based on functional dependence between intensity values of registered images. Although functional dependence is too restrictive on the global scale, earlier work has shown competitive performance in deformable registration when such measures are applied over small enough contexts. We confirm this finding and further develop the idea by modeling local functional dependence via the linear basis function model with the basis functions learned jointly with the deformation. The measure can be implemented via convolutions, making it efficient to compute on GPUs. We release the method as an easy-to-use tool and show good performance on three datasets compared to well-established baseline and earlier functional dependence-based methods.
☆ PhysicsGen: Can Generative Models Learn from Images to Predict Complex Physical Relations?
The image-to-image translation abilities of generative learning models have recently made significant progress in the estimation of complex (steered) mappings between image distributions. While appearance based tasks like image in-painting or style transfer have been studied at length, we propose to investigate the potential of generative models in the context of physical simulations. Providing a dataset of 300k image-pairs and baseline evaluations for three different physical simulation tasks, we propose a benchmark to investigate the following research questions: i) are generative models able to learn complex physical relations from input-output image pairs? ii) what speedups can be achieved by replacing differential equation based simulations? While baseline evaluations of different current models show the potential for high speedups (ii), these results also show strong limitations toward the physical correctness (i). This underlines the need for new methods to enforce physical correctness. Data, baseline models and evaluation code http://www.physics-gen.org.
☆ CoMoGaussian: Continuous Motion-Aware Gaussian Splatting from Motion-Blurred Images
3D Gaussian Splatting (3DGS) has gained significant attention for their high-quality novel view rendering, motivating research to address real-world challenges. A critical issue is the camera motion blur caused by movement during exposure, which hinders accurate 3D scene reconstruction. In this study, we propose CoMoGaussian, a Continuous Motion-Aware Gaussian Splatting that reconstructs precise 3D scenes from motion-blurred images while maintaining real-time rendering speed. Considering the complex motion patterns inherent in real-world camera movements, we predict continuous camera trajectories using neural ordinary differential equations (ODEs). To ensure accurate modeling, we employ rigid body transformations, preserving the shape and size of the object but rely on the discrete integration of sampled frames. To better approximate the continuous nature of motion blur, we introduce a continuous motion refinement (CMR) transformation that refines rigid transformations by incorporating additional learnable parameters. By revisiting fundamental camera theory and leveraging advanced neural ODE techniques, we achieve precise modeling of continuous camera trajectories, leading to improved reconstruction accuracy. Extensive experiments demonstrate state-of-the-art performance both quantitatively and qualitatively on benchmark datasets, which include a wide range of motion blur scenarios, from moderate to extreme blur.
comment: Revised Version of CRiM-GS, Github: https://github.com/Jho-Yonsei/CoMoGaussian
☆ Attenuation artifact detection and severity classification in intracoronary OCT using mixed image representations
In intracoronary optical coherence tomography (OCT), blood residues and gas bubbles cause attenuation artifacts that can obscure critical vessel structures. The presence and severity of these artifacts may warrant re-acquisition, prolonging procedure time and increasing use of contrast agent. Accurate detection of these artifacts can guide targeted re-acquisition, reducing the amount of repeated scans needed to achieve diagnostically viable images. However, the highly heterogeneous appearance of these artifacts poses a challenge for the automated detection of the affected image regions. To enable automatic detection of the attenuation artifacts caused by blood residues and gas bubbles based on their severity, we propose a convolutional neural network that performs classification of the attenuation lines (A-lines) into three classes: no artifact, mild artifact and severe artifact. Our model extracts and merges features from OCT images in both Cartesian and polar coordinates, where each column of the image represents an A-line. Our method detects the presence of attenuation artifacts in OCT frames reaching F-scores of 0.77 and 0.94 for mild and severe artifacts, respectively. The inference time over a full OCT scan is approximately 6 seconds. Our experiments show that analysis of images represented in both Cartesian and polar coordinate systems outperforms the analysis in polar coordinates only, suggesting that these representations contain complementary features. This work lays the foundation for automated artifact assessment and image acquisition guidance in intracoronary OCT imaging.
☆ Robust Multimodal Learning for Ophthalmic Disease Grading via Disentangled Representation
This paper discusses how ophthalmologists often rely on multimodal data to improve diagnostic accuracy. However, complete multimodal data is rare in real-world applications due to a lack of medical equipment and concerns about data privacy. Traditional deep learning methods typically address these issues by learning representations in latent space. However, the paper highlights two key limitations of these approaches: (i) Task-irrelevant redundant information (e.g., numerous slices) in complex modalities leads to significant redundancy in latent space representations. (ii) Overlapping multimodal representations make it difficult to extract unique features for each modality. To overcome these challenges, the authors propose the Essence-Point and Disentangle Representation Learning (EDRL) strategy, which integrates a self-distillation mechanism into an end-to-end framework to enhance feature selection and disentanglement for more robust multimodal learning. Specifically, the Essence-Point Representation Learning module selects discriminative features that improve disease grading performance. The Disentangled Representation Learning module separates multimodal data into modality-common and modality-unique representations, reducing feature entanglement and enhancing both robustness and interpretability in ophthalmic disease diagnosis. Experiments on multimodal ophthalmology datasets show that the proposed EDRL strategy significantly outperforms current state-of-the-art methods.
comment: 10pages
☆ Frequency Autoregressive Image Generation with Continuous Tokens
Autoregressive (AR) models for image generation typically adopt a two-stage paradigm of vector quantization and raster-scan ``next-token prediction", inspired by its great success in language modeling. However, due to the huge modality gap, image autoregressive models may require a systematic reevaluation from two perspectives: tokenizer format and regression direction. In this paper, we introduce the frequency progressive autoregressive (\textbf{FAR}) paradigm and instantiate FAR with the continuous tokenizer. Specifically, we identify spectral dependency as the desirable regression direction for FAR, wherein higher-frequency components build upon the lower one to progressively construct a complete image. This design seamlessly fits the causality requirement for autoregressive models and preserves the unique spatial locality of image data. Besides, we delve into the integration of FAR and the continuous tokenizer, introducing a series of techniques to address optimization challenges and improve the efficiency of training and inference processes. We demonstrate the efficacy of FAR through comprehensive experiments on the ImageNet dataset and verify its potential on text-to-image generation.
☆ Escaping Plato's Cave: Towards the Alignment of 3D and Text Latent Spaces CVPR 2025
Recent works have shown that, when trained at scale, uni-modal 2D vision and text encoders converge to learned features that share remarkable structural properties, despite arising from different representations. However, the role of 3D encoders with respect to other modalities remains unexplored. Furthermore, existing 3D foundation models that leverage large datasets are typically trained with explicit alignment objectives with respect to frozen encoders from other representations. In this work, we investigate the possibility of a posteriori alignment of representations obtained from uni-modal 3D encoders compared to text-based feature spaces. We show that naive post-training feature alignment of uni-modal text and 3D encoders results in limited performance. We then focus on extracting subspaces of the corresponding feature spaces and discover that by projecting learned representations onto well-chosen lower-dimensional subspaces the quality of alignment becomes significantly higher, leading to improved accuracy on matching and retrieval tasks. Our analysis further sheds light on the nature of these shared subspaces, which roughly separate between semantic and geometric data representations. Overall, ours is the first work that helps to establish a baseline for post-training alignment of 3D uni-modal and text feature spaces, and helps to highlight both the shared and unique properties of 3D data compared to other representations.
comment: Accepted at CVPR 2025
☆ CMMCoT: Enhancing Complex Multi-Image Comprehension via Multi-Modal Chain-of-Thought and Memory Augmentation
While previous multimodal slow-thinking methods have demonstrated remarkable success in single-image understanding scenarios, their effectiveness becomes fundamentally constrained when extended to more complex multi-image comprehension tasks. This limitation stems from their predominant reliance on text-based intermediate reasoning processes. While for human, when engaging in sophisticated multi-image analysis, they typically perform two complementary cognitive operations: (1) continuous cross-image visual comparison through region-of-interest matching, and (2) dynamic memorization of critical visual concepts throughout the reasoning chain. Motivated by these observations, we propose the Complex Multi-Modal Chain-of-Thought (CMMCoT) framework, a multi-step reasoning framework that mimics human-like "slow thinking" for multi-image understanding. Our approach incorporates two key innovations: 1. The construction of interleaved multimodal multi-step reasoning chains, which utilize critical visual region tokens, extracted from intermediate reasoning steps, as supervisory signals. This mechanism not only facilitates comprehensive cross-modal understanding but also enhances model interpretability. 2. The introduction of a test-time memory augmentation module that expands the model reasoning capacity during inference while preserving parameter efficiency. Furthermore, to facilitate research in this direction, we have curated a novel multi-image slow-thinking dataset. Extensive experiments demonstrate the effectiveness of our model.
☆ ColFigPhotoAttnNet: Reliable Finger Photo Presentation Attack Detection Leveraging Window-Attention on Color Spaces WACV
Finger photo Presentation Attack Detection (PAD) can significantly strengthen smartphone device security. However, these algorithms are trained to detect certain types of attacks. Furthermore, they are designed to operate on images acquired by specific capture devices, leading to poor generalization and a lack of robustness in handling the evolving nature of mobile hardware. The proposed investigation is the first to systematically analyze the performance degradation of existing deep learning PAD systems, convolutional and transformers, in cross-capture device settings. In this paper, we introduce the ColFigPhotoAttnNet architecture designed based on window attention on color channels, followed by the nested residual network as the predictor to achieve a reliable PAD. Extensive experiments using various capture devices, including iPhone13 Pro, GooglePixel 3, Nokia C5, and OnePlusOne, were carried out to evaluate the performance of proposed and existing methods on three publicly available databases. The findings underscore the effectiveness of our approach.
comment: Accepted in Winter Conference on Applications of Computer Vision (WACV) 2025
☆ L-FUSION: Laplacian Fetal Ultrasound Segmentation & Uncertainty Estimation
Accurate analysis of prenatal ultrasound (US) is essential for early detection of developmental anomalies. However, operator dependency and technical limitations (e.g. intrinsic artefacts and effects, setting errors) can complicate image interpretation and the assessment of diagnostic uncertainty. We present L-FUSION (Laplacian Fetal US Segmentation with Integrated FoundatiON models), a framework that integrates uncertainty quantification through unsupervised, normative learning and large-scale foundation models for robust segmentation of fetal structures in normal and pathological scans. We propose to utilise the aleatoric logit distributions of Stochastic Segmentation Networks and Laplace approximations with fast Hessian estimations to estimate epistemic uncertainty only from the segmentation head. This enables us to achieve reliable abnormality quantification for instant diagnostic feedback. Combined with an integrated Dropout component, L-FUSION enables reliable differentiation of lesions from normal fetal anatomy with enhanced uncertainty maps and segmentation counterfactuals in US imaging. It improves epistemic and aleatoric uncertainty interpretation and removes the need for manual disease-labelling. Evaluations across multiple datasets show that L-FUSION achieves superior segmentation accuracy and consistent uncertainty quantification, supporting on-site decision-making and offering a scalable solution for advancing fetal ultrasound analysis in clinical settings.
comment: Under Review
☆ Unified Reward Model for Multimodal Understanding and Generation
Recent advances in human preference alignment have significantly enhanced multimodal generation and understanding. A key approach is training reward models to guide preference optimization. However, existing models are often task-specific, limiting their adaptability across diverse visual applications. We also argue that jointly learning to assess multiple tasks may foster a synergistic effect, where improved image understanding enhances image generation assessment, and refined image evaluation benefits video assessment through better frame analysis. To this end, this paper proposes UnifiedReward, the first unified reward model for multimodal understanding and generation assessment, enabling both pairwise ranking and pointwise scoring, which can be employed for vision model preference alignment. Specifically, (1) we first develop UnifiedReward on our constructed large-scale human preference dataset, including both image and video generation/understanding tasks. (2) Then, it is utilized to automatically construct high-quality preference pair data based on the vision models, fine-gradually filtering their outputs through pair ranking and point sifting. (3) Finally, these data are used for their preference alignment through Direct Preference Optimization (DPO). Experimental results demonstrate that joint learning to assess diverse visual tasks can lead to substantial mutual benefits and we apply our pipeline to both image and video understanding/generation tasks, significantly improving the performance in each domain.
comment: project page: https://codegoat24.github.io/UnifiedReward/
☆ RecipeGen: A Benchmark for Real-World Recipe Image Generation
Recipe image generation is an important challenge in food computing, with applications from culinary education to interactive recipe platforms. However, there is currently no real-world dataset that comprehensively connects recipe goals, sequential steps, and corresponding images. To address this, we introduce RecipeGen, the first real-world goal-step-image benchmark for recipe generation, featuring diverse ingredients, varied recipe steps, multiple cooking styles, and a broad collection of food categories. Data is in https://github.com/zhangdaxia22/RecipeGen.
☆ DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker Characteristics And Intelligibility NAACL 25
Video-to-speech (V2S) synthesis, the task of generating speech directly from silent video input, is inherently more challenging than other speech synthesis tasks due to the need to accurately reconstruct both speech content and speaker characteristics from visual cues alone. Recently, audio-visual pre-training has eliminated the need for additional acoustic hints in V2S, which previous methods often relied on to ensure training convergence. However, even with pre-training, existing methods continue to face challenges in achieving a balance between acoustic intelligibility and the preservation of speaker-specific characteristics. We analyzed this limitation and were motivated to introduce DiVISe (Direct Visual-Input Speech Synthesis), an end-to-end V2S model that predicts Mel-spectrograms directly from video frames alone. Despite not taking any acoustic hints, DiVISe effectively preserves speaker characteristics in the generated audio, and achieves superior performance on both objective and subjective metrics across the LRS2 and LRS3 datasets. Our results demonstrate that DiVISe not only outperforms existing V2S models in acoustic intelligibility but also scales more effectively with increased data and model parameters. Code and weights can be found at https://github.com/PussyCat0700/DiVISe.
comment: to be published in NAACL 25
☆ Separability Membrane: 3D Active Contour for Point Cloud Surface Reconstruction
This paper proposes Separability Membrane, a robust 3D active contour for extracting a surface from 3D point cloud object. Our approach defines the surface of a 3D object as the boundary that maximizes the separability of point features, such as intensity, color, or local density, between its inner and outer regions based on Fisher's ratio. Separability Membrane identifies the exact surface of a 3D object by maximizing class separability while controlling the rigidity of the 3D surface model with an adaptive B-spline surface that adjusts its properties based on the local and global separability. A key advantage of our method is its ability to accurately reconstruct surface boundaries even when they are ambiguous due to noise or outliers, without requiring any training data or conversion to volumetric representation. Evaluations on a synthetic 3D point cloud dataset and the 3DNet dataset demonstrate the membrane's effectiveness and robustness under diverse conditions.
☆ Gaussian Random Fields as an Abstract Representation of Patient Metadata for Multimodal Medical Image Segmentation
The growing rate of chronic wound occurrence, especially in patients with diabetes, has become a concerning trend in recent years. Chronic wounds are difficult and costly to treat, and have become a serious burden on health care systems worldwide. Chronic wounds can have devastating consequences for the patient, with infection often leading to reduced quality of life and increased mortality risk. Innovative deep learning methods for the detection and monitoring of such wounds have the potential to reduce the impact to both patient and clinician. We present a novel multimodal segmentation method which allows for the introduction of patient metadata into the training workflow whereby the patient data are expressed as Gaussian random fields. Our results indicate that the proposed method improved performance when utilising multiple models, each trained on different metadata categories. Using the Diabetic Foot Ulcer Challenge 2022 test set, when compared to the baseline results (intersection over union = 0.4670, Dice similarity coefficient = 0.5908) we demonstrate improvements of +0.0220 and +0.0229 for intersection over union and Dice similarity coefficient respectively. This paper presents the first study to focus on integrating patient data into a chronic wound segmentation workflow. Our results show significant performance gains when training individual models using specific metadata categories, followed by average merging of prediction masks using distance transforms. All source code for this study is available at: https://github.com/mmu-dermatology-research/multimodal-grf
☆ Data-Efficient Generalization for Zero-shot Composed Image Retrieval
Zero-shot Composed Image Retrieval (ZS-CIR) aims to retrieve the target image based on a reference image and a text description without requiring in-distribution triplets for training. One prevalent approach follows the vision-language pretraining paradigm that employs a mapping network to transfer the image embedding to a pseudo-word token in the text embedding space. However, this approach tends to impede network generalization due to modality discrepancy and distribution shift between training and inference. To this end, we propose a Data-efficient Generalization (DeG) framework, including two novel designs, namely, Textual Supplement (TS) module and Semantic-Set (S-Set). The TS module exploits compositional textual semantics during training, enhancing the pseudo-word token with more linguistic semantics and thus mitigating the modality discrepancy effectively. The S-Set exploits the zero-shot capability of pretrained Vision-Language Models (VLMs), alleviating the distribution shift and mitigating the overfitting issue from the redundancy of the large-scale image-text data. Extensive experiments over four ZS-CIR benchmarks show that DeG outperforms the state-of-the-art (SOTA) methods with much less training data, and saves substantial training and inference time for practical usage.
☆ STGA: Selective-Training Gaussian Head Avatars
We propose selective-training Gaussian head avatars (STGA) to enhance the details of dynamic head Gaussian. The dynamic head Gaussian model is trained based on the FLAME parameterized model. Each Gaussian splat is embedded within the FLAME mesh to achieve mesh-based animation of the Gaussian model. Before training, our selection strategy calculates the 3D Gaussian splat to be optimized in each frame. The parameters of these 3D Gaussian splats are optimized in the training of each frame, while those of the other splats are frozen. This means that the splats participating in the optimization process differ in each frame, to improve the realism of fine details. Compared with network-based methods, our method achieves better results with shorter training time. Compared with mesh-based methods, our method produces more realistic details within the same training time. Additionally, the ablation experiment confirms that our method effectively enhances the quality of details.
☆ Partially Supervised Unpaired Multi-Modal Learning for Label-Efficient Medical Image Segmentation
Unpaired Multi-Modal Learning (UMML) which leverages unpaired multi-modal data to boost model performance on each individual modality has attracted a lot of research interests in medical image analysis. However, existing UMML methods require multi-modal datasets to be fully labeled, which incurs tremendous annotation cost. In this paper, we investigate the use of partially labeled data for label-efficient unpaired multi-modal learning, which can reduce the annotation cost by up to one half. We term the new learning paradigm as Partially Supervised Unpaired Multi-Modal Learning (PSUMML) and propose a novel Decomposed partial class adaptation with snapshot Ensembled Self-Training (DEST) framework for it. Specifically, our framework consists of a compact segmentation network with modality specific normalization layers for learning with partially labeled unpaired multi-modal data. The key challenge in PSUMML lies in the complex partial class distribution discrepancy due to partial class annotation, which hinders effective knowledge transfer across modalities. We theoretically analyze this phenomenon with a decomposition theorem and propose a decomposed partial class adaptation technique to precisely align the partially labeled classes across modalities to reduce the distribution discrepancy. We further propose a snapshot ensembled self-training technique to leverage the valuable snapshot models during training to assign pseudo-labels to partially labeled pixels for self-training to boost model performance. We perform extensive experiments under different scenarios of PSUMML for two medical image segmentation tasks, namely cardiac substructure segmentation and abdominal multi-organ segmentation. Our framework outperforms existing methods significantly.
comment: Accepted to MLMI 2024
☆ Narrating the Video: Boosting Text-Video Retrieval via Comprehensive Utilization of Frame-Level Captions CVPR 2025
In recent text-video retrieval, the use of additional captions from vision-language models has shown promising effects on the performance. However, existing models using additional captions often have struggled to capture the rich semantics, including temporal changes, inherent in the video. In addition, incorrect information caused by generative models can lead to inaccurate retrieval. To address these issues, we propose a new framework, Narrating the Video (NarVid), which strategically leverages the comprehensive information available from frame-level captions, the narration. The proposed NarVid exploits narration in multiple ways: 1) feature enhancement through cross-modal interactions between narration and video, 2) query-aware adaptive filtering to suppress irrelevant or incorrect information, 3) dual-modal matching score by adding query-video similarity and query-narration similarity, and 4) hard-negative loss to learn discriminative features from multiple perspectives using the two similarities from different views. Experimental results demonstrate that NarVid achieves state-of-the-art performance on various benchmark datasets.
comment: Accepted at CVPR 2025
☆ Spectral-Spatial Extraction through Layered Tensor Decomposition for Hyperspectral Anomaly Detection
Low rank tensor representation (LRTR) methods are very useful for hyperspectral anomaly detection (HAD). To overcome the limitations that they often overlook spectral anomaly and rely on large-scale matrix singular value decomposition, we first apply non-negative matrix factorization (NMF) to alleviate spectral dimensionality redundancy and extract spectral anomaly and then employ LRTR to extract spatial anomaly while mitigating spatial redundancy, yielding a highly efffcient layered tensor decomposition (LTD) framework for HAD. An iterative algorithm based on proximal alternating minimization is developed to solve the proposed LTD model, with convergence guarantees provided. Moreover, we introduce a rank reduction strategy with validation mechanism that adaptively reduces data size while preventing excessive reduction. Theoretically, we rigorously establish the equivalence between the tensor tubal rank and tensor group sparsity regularization (TGSR) and, under mild conditions, demonstrate that the relaxed formulation of TGSR shares the same global minimizers and optimal values as its original counterpart. Experimental results on the Airport-Beach-Urban and MVTec datasets demonstrate that our approach outperforms state-of-the-art methods in the HAD task.
☆ MGSR: 2D/3D Mutual-boosted Gaussian Splatting for High-fidelity Surface Reconstruction under Various Light Conditions
Novel view synthesis (NVS) and surface reconstruction (SR) are essential tasks in 3D Gaussian Splatting (3D-GS). Despite recent progress, these tasks are often addressed independently, with GS-based rendering methods struggling under diverse light conditions and failing to produce accurate surfaces, while GS-based reconstruction methods frequently compromise rendering quality. This raises a central question: must rendering and reconstruction always involve a trade-off? To address this, we propose MGSR, a 2D/3D Mutual-boosted Gaussian splatting for Surface Reconstruction that enhances both rendering quality and 3D reconstruction accuracy. MGSR introduces two branches--one based on 2D-GS and the other on 3D-GS. The 2D-GS branch excels in surface reconstruction, providing precise geometry information to the 3D-GS branch. Leveraging this geometry, the 3D-GS branch employs a geometry-guided illumination decomposition module that captures reflected and transmitted components, enabling realistic rendering under varied light conditions. Using the transmitted component as supervision, the 2D-GS branch also achieves high-fidelity surface reconstruction. Throughout the optimization process, the 2D-GS and 3D-GS branches undergo alternating optimization, providing mutual supervision. Prior to this, each branch completes an independent warm-up phase, with an early stopping strategy implemented to reduce computational costs. We evaluate MGSR on a diverse set of synthetic and real-world datasets, at both object and scene levels, demonstrating strong performance in rendering and surface reconstruction.
comment: 11 pages, 7 figures
☆ SplatPose: Geometry-Aware 6-DoF Pose Estimation from Single RGB Image via 3D Gaussian Splatting IROS 2025
6-DoF pose estimation is a fundamental task in computer vision with wide-ranging applications in augmented reality and robotics. Existing single RGB-based methods often compromise accuracy due to their reliance on initial pose estimates and susceptibility to rotational ambiguity, while approaches requiring depth sensors or multi-view setups incur significant deployment costs. To address these limitations, we introduce SplatPose, a novel framework that synergizes 3D Gaussian Splatting (3DGS) with a dual-branch neural architecture to achieve high-precision pose estimation using only a single RGB image. Central to our approach is the Dual-Attention Ray Scoring Network (DARS-Net), which innovatively decouples positional and angular alignment through geometry-domain attention mechanisms, explicitly modeling directional dependencies to mitigate rotational ambiguity. Additionally, a coarse-to-fine optimization pipeline progressively refines pose estimates by aligning dense 2D features between query images and 3DGS-synthesized views, effectively correcting feature misalignment and depth errors from sparse ray sampling. Experiments on three benchmark datasets demonstrate that SplatPose achieves state-of-the-art 6-DoF pose estimation accuracy in single RGB settings, rivaling approaches that depend on depth or multi-view images.
comment: Submitted to IROS 2025
☆ Spatial Context-Driven Positive Pair Sampling for Enhanced Histopathology Image Classification
Deep learning has demonstrated great promise in cancer classification from whole-slide images (WSIs) but remains constrained by the need for extensive annotations. Annotation-free methods, such as multiple instance learning (MIL) and self-supervised learning (SSL), have emerged to address this challenge; however, current SSL techniques often depend on synthetic augmentations or temporal context, which may not adequately capture the intricate spatial relationships inherent to histopathology. In this work, we introduce a novel spatial context-driven positive pair sampling strategy for SSL that leverages the natural coherence of adjacent patches in WSIs. By constructing biologically relevant positive pairs from spatially proximate patches, our approach harnesses inherent spatial coherence to enhance patch-level representations, ultimately boosting slide-level classification performance. Experiments on multiple datasets reveal that our strategy improves classification accuracy by 5\% to 10\% over the standard method, paving the way for more clinically relevant AI models in cancer diagnosis. The code is available at https://anonymous.4open.science/r/contextual-pairs-E72F/.
☆ EvolvingGS: High-Fidelity Streamable Volumetric Video via Evolving 3D Gaussian Representation
We have recently seen great progress in 3D scene reconstruction through explicit point-based 3D Gaussian Splatting (3DGS), notable for its high quality and fast rendering speed. However, reconstructing dynamic scenes such as complex human performances with long durations remains challenging. Prior efforts fall short of modeling a long-term sequence with drastic motions, frequent topology changes or interactions with props, and resort to segmenting the whole sequence into groups of frames that are processed independently, which undermines temporal stability and thereby leads to an unpleasant viewing experience and inefficient storage footprint. In view of this, we introduce EvolvingGS, a two-stage strategy that first deforms the Gaussian model to coarsely align with the target frame, and then refines it with minimal point addition/subtraction, particularly in fast-changing areas. Owing to the flexibility of the incrementally evolving representation, our method outperforms existing approaches in terms of both per-frame and temporal quality metrics while maintaining fast rendering through its purely explicit representation. Moreover, by exploiting temporal coherence between successive frames, we propose a simple yet effective compression algorithm that achieves over 50x compression rate. Extensive experiments on both public benchmarks and challenging custom datasets demonstrate that our method significantly advances the state-of-the-art in dynamic scene reconstruction, particularly for extended sequences with complex human performances.
☆ GaussianCAD: Robust Self-Supervised CAD Reconstruction from Three Orthographic Views Using 3D Gaussian Splatting
The automatic reconstruction of 3D computer-aided design (CAD) models from CAD sketches has recently gained significant attention in the computer vision community. Most existing methods, however, rely on vector CAD sketches and 3D ground truth for supervision, which are often difficult to be obtained in industrial applications and are sensitive to noise inputs. We propose viewing CAD reconstruction as a specific instance of sparse-view 3D reconstruction to overcome these limitations. While this reformulation offers a promising perspective, existing 3D reconstruction methods typically require natural images and corresponding camera poses as inputs, which introduces two major significant challenges: (1) modality discrepancy between CAD sketches and natural images, and (2) difficulty of accurate camera pose estimation for CAD sketches. To solve these issues, we first transform the CAD sketches into representations resembling natural images and extract corresponding masks. Next, we manually calculate the camera poses for the orthographic views to ensure accurate alignment within the 3D coordinate system. Finally, we employ a customized sparse-view 3D reconstruction method to achieve high-quality reconstructions from aligned orthographic views. By leveraging raster CAD sketches for self-supervision, our approach eliminates the reliance on vector CAD sketches and 3D ground truth. Experiments on the Sub-Fusion360 dataset demonstrate that our proposed method significantly outperforms previous approaches in CAD reconstruction performance and exhibits strong robustness to noisy inputs.
☆ Accelerating Diffusion Transformer via Gradient-Optimized Cache
Feature caching has emerged as an effective strategy to accelerate diffusion transformer (DiT) sampling through temporal feature reuse. It is a challenging problem since (1) Progressive error accumulation from cached blocks significantly degrades generation quality, particularly when over 50\% of blocks are cached; (2) Current error compensation approaches neglect dynamic perturbation patterns during the caching process, leading to suboptimal error correction. To solve these problems, we propose the Gradient-Optimized Cache (GOC) with two key innovations: (1) Cached Gradient Propagation: A gradient queue dynamically computes the gradient differences between cached and recomputed features. These gradients are weighted and propagated to subsequent steps, directly compensating for the approximation errors introduced by caching. (2) Inflection-Aware Optimization: Through statistical analysis of feature variation patterns, we identify critical inflection points where the denoising trajectory changes direction. By aligning gradient updates with these detected phases, we prevent conflicting gradient directions during error correction. Extensive evaluations on ImageNet demonstrate GOC's superior trade-off between efficiency and quality. With 50\% cached blocks, GOC achieves IS 216.28 (26.3\% higher) and FID 3.907 (43\% lower) compared to baseline DiT, while maintaining identical computational costs. These improvements persist across various cache ratios, demonstrating robust adaptability to different acceleration requirements.
☆ Development and Enhancement of Text-to-Image Diffusion Models
This research focuses on the development and enhancement of text-to-image denoising diffusion models, addressing key challenges such as limited sample diversity and training instability. By incorporating Classifier-Free Guidance (CFG) and Exponential Moving Average (EMA) techniques, this study significantly improves image quality, diversity, and stability. Utilizing Hugging Face's state-of-the-art text-to-image generation model, the proposed enhancements establish new benchmarks in generative AI. This work explores the underlying principles of diffusion models, implements advanced strategies to overcome existing limitations, and presents a comprehensive evaluation of the improvements achieved. Results demonstrate substantial progress in generating stable, diverse, and high-quality images from textual descriptions, advancing the field of generative artificial intelligence and providing new foundations for future applications. Keywords: Text-to-image, Diffusion model, Classifier-free guidance, Exponential moving average, Image generation.
☆ R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model
Recently DeepSeek R1 demonstrated how reinforcement learning with simple rule-based incentives can enable autonomous development of complex reasoning in large language models, characterized by the "aha moment", in which the model manifest self-reflection and increased response length during training. However, attempts to extend this success to multimodal reasoning often failed to reproduce these key characteristics. In this report, we present the first successful replication of these emergent characteristics for multimodal reasoning on only a non-SFT 2B model. Starting with Qwen2-VL-2B and applying reinforcement learning directly on the SAT dataset, our model achieves 59.47% accuracy on CVBench, outperforming the base model by approximately ~30% and exceeding both SFT setting by ~2%. In addition, we share our failed attempts and insights in attempting to achieve R1-like reasoning using RL with instruct models. aiming to shed light on the challenges involved. Our key observations include: (1) applying RL on instruct model often results in trivial reasoning trajectories, and (2) naive length reward are ineffective in eliciting reasoning capabilities. The project code is available at https://github.com/turningpoint-ai/VisualThinker-R1-Zero
comment: 10 pages, 6 figures
☆ HexPlane Representation for 3D Semantic Scene Understanding
In this paper, we introduce the HexPlane representation for 3D semantic scene understanding. Specifically, we first design the View Projection Module (VPM) to project the 3D point cloud into six planes to maximally retain the original spatial information. Features of six planes are extracted by the 2D encoder and sent to the HexPlane Association Module (HAM) to adaptively fuse the most informative information for each point. The fused point features are further fed to the task head to yield the ultimate predictions. Compared to the popular point and voxel representation, the HexPlane representation is efficient and can utilize highly optimized 2D operations to process sparse and unordered 3D point clouds. It can also leverage off-the-shelf 2D models, network weights, and training recipes to achieve accurate scene understanding in 3D space. On ScanNet and SemanticKITTI benchmarks, our algorithm, dubbed HexNet3D, achieves competitive performance with previous algorithms. In particular, on the ScanNet 3D segmentation task, our method obtains 77.0 mIoU on the validation set, surpassing Point Transformer V2 by 1.6 mIoU. We also observe encouraging results in indoor 3D detection tasks. Note that our method can be seamlessly integrated into existing voxel-based, point-based, and range-based approaches and brings considerable gains without bells and whistles. The codes will be available upon publication.
comment: 7 pages, 2 figures
☆ EDM: Efficient Deep Feature Matching
Recent feature matching methods have achieved remarkable performance but lack efficiency consideration. In this paper, we revisit the mainstream detector-free matching pipeline and improve all its stages considering both accuracy and efficiency. We propose an Efficient Deep feature Matching network, EDM. We first adopt a deeper CNN with fewer dimensions to extract multi-level features. Then we present a Correlation Injection Module that conducts feature transformation on high-level deep features, and progressively injects feature correlations from global to local for efficient multi-scale feature aggregation, improving both speed and performance. In the refinement stage, a novel lightweight bidirectional axis-based regression head is designed to directly predict subpixel-level correspondences from latent features, avoiding the significant computational cost of explicitly locating keypoints on high-resolution local feature heatmaps. Moreover, effective selection strategies are introduced to enhance matching accuracy. Extensive experiments show that our EDM achieves competitive matching accuracy on various benchmarks and exhibits excellent efficiency, offering valuable best practices for real-world applications. The code is available at https://github.com/chicleee/EDM.
☆ SMILENet: Unleashing Extra-Large Capacity Image Steganography via a Synergistic Mosaic InvertibLE Hiding Network
Existing image steganography methods face fundamental limitations in hiding capacity (typically $1\sim7$ images) due to severe information interference and uncoordinated capacity-distortion trade-off. We propose SMILENet, a novel synergistic framework that achieves 25 image hiding through three key innovations: (i) A synergistic network architecture coordinates reversible and non-reversible operations to efficiently exploit information redundancy in both secret and cover images. The reversible Invertible Cover-Driven Mosaic (ICDM) module and Invertible Mosaic Secret Embedding (IMSE) module establish cover-guided mosaic transformations and representation embedding with mathematically guaranteed invertibility for distortion-free embedding. The non-reversible Secret Information Selection (SIS) module and Secret Detail Enhancement (SDE) module implement learnable feature modulation for critical information selection and enhancement. (ii) A unified training strategy that coordinates complementary modules to achieve 3.0x higher capacity than existing methods with superior visual quality. (iii) Last but not least, we introduce a new metric to model Capacity-Distortion Trade-off for evaluating the image steganography algorithms that jointly considers hiding capacity and distortion, and provides a unified evaluation approach for accessing results with different number of secret image. Extensive experiments on DIV2K, Paris StreetView and ImageNet1K show that SMILENet outperforms state-of-the-art methods in terms of hiding capacity, recovery quality as well as security against steganalysis methods.
☆ We Care Each Pixel: Calibrating on Medical Segmentation Model
Medical image segmentation is fundamental for computer-aided diagnostics, providing accurate delineation of anatomical structures and pathological regions. While common metrics such as Accuracy, DSC, IoU, and HD primarily quantify spatial agreement between predictions and ground-truth labels, they do not assess the calibration quality of segmentation models, which is crucial for clinical reliability. To address this limitation, we propose pixel-wise Expected Calibration Error (pECE), a novel metric that explicitly measures miscalibration at the pixel level, thereby ensuring both spatial precision and confidence reliability. We further introduce a morphological adaptation strategy that applies morphological operations to ground-truth masks before computing calibration losses, particularly benefiting margin-based losses such as Margin SVLS and NACL. Additionally, we present the Signed Distance Calibration Loss (SDC), which aligns boundary geometry with calibration objectives by penalizing discrepancies between predicted and ground-truth signed distance functions (SDFs). Extensive experiments demonstrate that our method not only enhances segmentation performance but also improves calibration quality, yielding more trustworthy confidence estimates. Code is available at: https://github.com/EagleAdelaide/SDC-Loss.
comment: Under Reviewing
☆ Visual Cues of Gender and Race are Associated with Stereotyping in Vision-Language Models
Current research on bias in Vision Language Models (VLMs) has important limitations: it is focused exclusively on trait associations while ignoring other forms of stereotyping, it examines specific contexts where biases are expected to appear, and it conceptualizes social categories like race and gender as binary, ignoring the multifaceted nature of these identities. Using standardized facial images that vary in prototypicality, we test four VLMs for both trait associations and homogeneity bias in open-ended contexts. We find that VLMs consistently generate more uniform stories for women compared to men, with people who are more gender prototypical in appearance being represented more uniformly. By contrast, VLMs represent White Americans more uniformly than Black Americans. Unlike with gender prototypicality, race prototypicality was not related to stronger uniformity. In terms of trait associations, we find limited evidence of stereotyping-Black Americans were consistently linked with basketball across all models, while other racial associations (i.e., art, healthcare, appearance) varied by specific VLM. These findings demonstrate that VLM stereotyping manifests in ways that go beyond simple group membership, suggesting that conventional bias mitigation strategies may be insufficient to address VLM stereotyping and that homogeneity bias persists even when trait associations are less apparent in model outputs.
☆ Fake It To Make It: Virtual Multiviews to Enhance Monocular Indoor Semantic Scene Completion IROS 2025
Monocular Indoor Semantic Scene Completion (SSC) aims to reconstruct a 3D semantic occupancy map from a single RGB image of an indoor scene, inferring spatial layout and object categories from 2D image cues. The challenge of this task arises from the depth, scale, and shape ambiguities that emerge when transforming a 2D image into 3D space, particularly within the complex and often heavily occluded environments of indoor scenes. Current SSC methods often struggle with these ambiguities, resulting in distorted or missing object representations. To overcome these limitations, we introduce an innovative approach that leverages novel view synthesis and multiview fusion. Specifically, we demonstrate how virtual cameras can be placed around the scene to emulate multiview inputs that enhance contextual scene information. We also introduce a Multiview Fusion Adaptor (MVFA) to effectively combine the multiview 3D scene predictions into a unified 3D semantic occupancy map. Finally, we identify and study the inherent limitation of generative techniques when applied to SSC, specifically the Novelty-Consistency tradeoff. Our system, GenFuSE, demonstrates IoU score improvements of up to 2.8% for Scene Completion and 4.9% for Semantic Scene Completion when integrated with existing SSC networks on the NYUv2 dataset. This work introduces GenFuSE as a standard framework for advancing monocular SSC with synthesized inputs.
comment: Submitted to IROS 2025
☆ Taming Video Diffusion Prior with Scene-Grounding Guidance for 3D Gaussian Splatting from Sparse Inputs CVPR2025
Despite recent successes in novel view synthesis using 3D Gaussian Splatting (3DGS), modeling scenes with sparse inputs remains a challenge. In this work, we address two critical yet overlooked issues in real-world sparse-input modeling: extrapolation and occlusion. To tackle these issues, we propose to use a reconstruction by generation pipeline that leverages learned priors from video diffusion models to provide plausible interpretations for regions outside the field of view or occluded. However, the generated sequences exhibit inconsistencies that do not fully benefit subsequent 3DGS modeling. To address the challenge of inconsistencies, we introduce a novel scene-grounding guidance based on rendered sequences from an optimized 3DGS, which tames the diffusion model to generate consistent sequences. This guidance is training-free and does not require any fine-tuning of the diffusion model. To facilitate holistic scene modeling, we also propose a trajectory initialization method. It effectively identifies regions that are outside the field of view and occluded. We further design a scheme tailored for 3DGS optimization with generated sequences. Experiments demonstrate that our method significantly improves upon the baseline and achieves state-of-the-art performance on challenging benchmarks.
comment: Accepted by CVPR2025. The project page is available at https://zhongyingji.github.io/guidevd-3dgs/
☆ Lightweight Hypercomplex MRI Reconstruction: A Generalized Kronecker-Parameterized Approach
Magnetic Resonance Imaging (MRI) is crucial for clinical diagnostics but is hindered by prolonged scan times. Current deep learning models enhance MRI reconstruction but are often memory-intensive and unsuitable for resource-limited systems. This paper introduces a lightweight MRI reconstruction model leveraging Kronecker-Parameterized Hypercomplex Neural Networks to achieve high performance with reduced parameters. By integrating Kronecker-based modules, including Kronecker MLP, Kronecker Window Attention, and Kronecker Convolution, the proposed model efficiently extracts spatial features while preserving representational power. We introduce Kronecker U-Net and Kronecker SwinMR, which maintain high reconstruction quality with approximately 50% fewer parameters compared to existing models. Experimental evaluation on the FastMRI dataset demonstrates competitive PSNR, SSIM, and LPIPS metrics, even at high acceleration factors (8x and 16x), with no significant performance drop. Additionally, Kronecker variants exhibit superior generalization and reduced overfitting on limited datasets, facilitating efficient MRI reconstruction on hardware-constrained systems. This approach sets a new benchmark for parameter-efficient medical imaging models.
comment: 11 pages, 3 figures. Submitted for publication
☆ Accelerated Patient-specific Non-Cartesian MRI Reconstruction using Implicit Neural Representations
The scanning time for a fully sampled MRI can be undesirably lengthy. Compressed sensing has been developed to minimize image artifacts in accelerated scans, but the required iterative reconstruction is computationally complex and difficult to generalize on new cases. Image-domain-based deep learning methods (e.g., convolutional neural networks) emerged as a faster alternative but face challenges in modeling continuous k-space, a problem amplified with non-Cartesian sampling commonly used in accelerated acquisition. In comparison, implicit neural representations can model continuous signals in the frequency domain and thus are compatible with arbitrary k-space sampling patterns. The current study develops a novel generative-adversarially trained implicit neural representations (k-GINR) for de novo undersampled non-Cartesian k-space reconstruction. k-GINR consists of two stages: 1) supervised training on an existing patient cohort; 2) self-supervised patient-specific optimization. In stage 1, the network is trained with the generative-adversarial network on diverse patients of the same anatomical region supervised by fully sampled acquisition. In stage 2, undersampled k-space data of individual patients is used to tailor the prior-embedded network for patient-specific optimization. The UCSF StarVIBE T1-weighted liver dataset was evaluated on the proposed framework. k-GINR is compared with an image-domain deep learning method, Deep Cascade CNN, and a compressed sensing method. k-GINR consistently outperformed the baselines with a larger performance advantage observed at very high accelerations (e.g., 20 times). k-GINR offers great value for direct non-Cartesian k-space reconstruction for new incoming patients across a wide range of accelerations liver anatomy.
☆ D2GV: Deformable 2D Gaussian Splatting for Video Representation in 400FPS
Implicit Neural Representations (INRs) have emerged as a powerful approach for video representation, offering versatility across tasks such as compression and inpainting. However, their implicit formulation limits both interpretability and efficacy, undermining their practicality as a comprehensive solution. We propose a novel video representation based on deformable 2D Gaussian splatting, dubbed D2GV, which aims to achieve three key objectives: 1) improved efficiency while delivering superior quality; 2) enhanced scalability and interpretability; and 3) increased friendliness for downstream tasks. Specifically, we initially divide the video sequence into fixed-length Groups of Pictures (GoP) to allow parallel training and linear scalability with video length. For each GoP, D2GV represents video frames by applying differentiable rasterization to 2D Gaussians, which are deformed from a canonical space into their corresponding timestamps. Notably, leveraging efficient CUDA-based rasterization, D2GV converges fast and decodes at speeds exceeding 400 FPS, while delivering quality that matches or surpasses state-of-the-art INRs. Moreover, we incorporate a learnable pruning and quantization strategy to streamline D2GV into a more compact representation. We demonstrate D2GV's versatility in tasks including video interpolation, inpainting and denoising, underscoring its potential as a promising solution for video representation. Code is available at: https://github.com/Evan-sudo/D2GV.
♻ ☆ Spatial regularisation for improved accuracy and interpretability in keypoint-based registration
Unsupervised registration strategies bypass requirements in ground truth transforms or segmentations by optimising similarity metrics between fixed and moved volumes. Among these methods, a recent subclass of approaches based on unsupervised keypoint detection stand out as very promising for interpretability. Specifically, these methods train a network to predict feature maps for fixed and moving images, from which explainable centres of mass are computed to obtain point clouds, that are then aligned in closed-form. However, the features returned by the network often yield spatially diffuse patterns that are hard to interpret, thus undermining the purpose of keypoint-based registration. Here, we propose a three-fold loss to regularise the spatial distribution of the features. First, we use the KL divergence to model features as point spread functions that we interpret as probabilistic keypoints. Then, we sharpen the spatial distributions of these features to increase the precision of the detected landmarks. Finally, we introduce a new repulsive loss across keypoints to encourage spatial diversity. Overall, our loss considerably improves the interpretability of the features, which now correspond to precise and anatomically meaningful landmarks. We demonstrate our three-fold loss in foetal rigid motion tracking and brain MRI affine registration tasks, where it not only outperforms state-of-the-art unsupervised strategies, but also bridges the gap with state-of-the-art supervised methods. Our code is available at https://github.com/BenBillot/spatial_regularisation.
comment: under review
♻ ☆ Question-Aware Gaussian Experts for Audio-Visual Question Answering CVPR 2025
Audio-Visual Question Answering (AVQA) requires not only question-based multimodal reasoning but also precise temporal grounding to capture subtle dynamics for accurate prediction. However, existing methods mainly use question information implicitly, limiting focus on question-specific details. Furthermore, most studies rely on uniform frame sampling, which can miss key question-relevant frames. Although recent Top-K frame selection methods aim to address this, their discrete nature still overlooks fine-grained temporal details. This paper proposes QA-TIGER, a novel framework that explicitly incorporates question information and models continuous temporal dynamics. Our key idea is to use Gaussian-based modeling to adaptively focus on both consecutive and non-consecutive frames based on the question, while explicitly injecting question information and applying progressive refinement. We leverage a Mixture of Experts (MoE) to flexibly implement multiple Gaussian models, activating temporal experts specifically tailored to the question. Extensive experiments on multiple AVQA benchmarks show that QA-TIGER consistently achieves state-of-the-art performance. Code is available at https://aim-skku.github.io/QA-TIGER/
comment: CVPR 2025. Code is available at https://github.com/AIM-SKKU/QA-TIGER
♻ ☆ LEDiT: Your Length-Extrapolatable Diffusion Transformer without Positional Encoding
Diffusion transformers(DiTs) struggle to generate images at resolutions higher than their training resolutions. The primary obstacle is that the explicit positional encodings(PE), such as RoPE, need extrapolation which degrades performance when the inference resolution differs from training. In this paper, we propose a Length-Extrapolatable Diffusion Transformer(LEDiT), a simple yet powerful architecture to overcome this limitation. LEDiT needs no explicit PEs, thereby avoiding extrapolation. The key innovations of LEDiT are introducing causal attention to implicitly impart global positional information to tokens, while enhancing locality to precisely distinguish adjacent tokens. Experiments on 256x256 and 512x512 ImageNet show that LEDiT can scale the inference resolution to 512x512 and 1024x1024, respectively, while achieving better image quality compared to current state-of-the-art length extrapolation methods(NTK-aware, YaRN). Moreover, LEDiT achieves strong extrapolation performance with just 100K steps of fine-tuning on a pretrained DiT, demonstrating its potential for integration into existing text-to-image DiTs. Project page: https://shenzhang2145.github.io/ledit/
♻ ☆ GBT-SAM: A Parameter-Efficient Depth-Aware Model for Generalizable Brain tumour Segmentation on mp-MRI
Gliomas are brain tumours that stand out for their highly lethal and aggressive nature, which demands a precise approach in their diagnosis. Medical image segmentation plays a crucial role in the evaluation and follow-up of these tumours, allowing specialists to analyse their morphology. However, existing methods for automatic glioma segmentation often lack generalization capability across other brain tumour domains, require extensive computational resources, or fail to fully utilize the multi-parametric MRI (mp-MRI) data used to delineate them. In this work, we introduce GBT-SAM, a novel Generalizable Brain Tumour (GBT) framework that extends the Segment Anything Model (SAM) to brain tumour segmentation tasks. Our method employs a two-step training protocol: first, fine-tuning the patch embedding layer to process the entire mp-MRI modalities, and second, incorporating parameter-efficient LoRA blocks and a Depth-Condition block into the Vision Transformer (ViT) to capture inter-slice correlations. GBT-SAM achieves state-of-the-art performance on the Adult Glioma dataset (Dice Score of $93.54$) while demonstrating robust generalization across Meningioma, Pediatric Glioma, and Sub-Saharan Glioma datasets. Furthermore, GBT-SAM uses less than 6.5M trainable parameters, thus offering an efficient solution for brain tumour segmentation. \\ Our code and models are available at https://github.com/vpulab/med-sam-brain .
♻ ☆ VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models
In this paper, we propose a novel framework for solving high-definition video inverse problems using latent image diffusion models. Building on recent advancements in spatio-temporal optimization for video inverse problems using image diffusion models, our approach leverages latent-space diffusion models to achieve enhanced video quality and resolution. To address the high computational demands of processing high-resolution frames, we introduce a pseudo-batch consistent sampling strategy, allowing efficient operation on a single GPU. Additionally, to improve temporal consistency, we present pseudo-batch inversion, an initialization technique that incorporates informative latents from the measurement. By integrating with SDXL, our framework achieves state-of-the-art video reconstruction across a wide range of spatio-temporal inverse problems, including complex combinations of frame averaging and various spatial degradations, such as deblurring, super-resolution, and inpainting. Unlike previous methods, our approach supports multiple aspect ratios (landscape, vertical, and square) and delivers HD-resolution reconstructions (exceeding 1280x720) in under 6 seconds per frame on a single NVIDIA 4090 GPU.
comment: Project page: https://vision-xl.github.io/
♻ ☆ Diff-Reg v2: Diffusion-Based Matching Matrix Estimation for Image Matching and 3D Registration
Establishing reliable correspondences is crucial for all registration tasks, including 2D image registration, 3D point cloud registration, and 2D-3D image-to-point cloud registration. However, these tasks are often complicated by challenges such as scale inconsistencies, symmetry, and large deformations, which can lead to ambiguous matches. Previous feature-based and correspondence-based methods typically rely on geometric or semantic features to generate or polish initial potential correspondences. Some methods typically leverage specific geometric priors, such as topological preservation, to devise diverse and innovative strategies tailored to a given enhancement goal, which cannot be exhaustively enumerated. Additionally, many previous approaches rely on a single-step prediction head, which can struggle with local minima in complex matching scenarios. To address these challenges, we introduce an innovative paradigm that leverages a diffusion model in matrix space for robust matching matrix estimation. Our model treats correspondence estimation as a denoising diffusion process in the matching matrix space, gradually refining the intermediate matching matrix to the optimal one. Specifically, we apply the diffusion model in the doubly stochastic matrix space for 3D-3D and 2D-3D registration tasks. In the 2D image registration task, we deploy the diffusion model in a matrix subspace where dual-softmax projection regularization is applied. For all three registration tasks, we provide adaptive matching matrix embedding implementations tailored to the specific characteristics of each task while maintaining a consistent "match-to-warp" encoding pattern. Furthermore, we adopt a lightweight design for the denoising module. In inference, once points or image features are extracted and fixed, this module performs multi-step denoising predictions through reverse sampling.
comment: arXiv admin note: text overlap with arXiv:2403.19919
♻ ☆ Tell Me What to Track: Infusing Robust Language Guidance for Enhanced Referring Multi-Object Tracking
Referring multi-object tracking (RMOT) is an emerging cross-modal task that aims to localize an arbitrary number of targets based on a language expression and continuously track them in a video. This intricate task involves reasoning on multi-modal data and precise target localization with temporal association. However, prior studies overlook the imbalanced data distribution between newborn targets and existing targets due to the nature of the task. In addition, they only indirectly fuse multi-modal features, struggling to deliver clear guidance on newborn target detection. To solve the above issues, we conduct a collaborative matching strategy to alleviate the impact of the imbalance, boosting the ability to detect newborn targets while maintaining tracking performance. In the encoder, we integrate and enhance the cross-modal and multi-scale fusion, overcoming the bottlenecks in previous work, where limited multi-modal information is shared and interacted between feature maps. In the decoder, we also develop a referring-infused adaptation that provides explicit referring guidance through the query tokens. The experiments showcase the superior performance of our model (+3.42%) compared to prior works, demonstrating the effectiveness of our designs.
♻ ☆ Motion by Queries: Identity-Motion Trade-offs in Text-to-Video Generation
Text-to-video diffusion models have shown remarkable progress in generating coherent video clips from textual descriptions. However, the interplay between motion, structure, and identity representations in these models remains under-explored. Here, we investigate how self-attention query features (a.k.a. Q features) simultaneously govern motion, structure, and identity and examine the challenges arising when these representations interact. Our analysis reveals that Q affects not only layout, but that during denoising Q also has a strong effect on subject identity, making it hard to transfer motion without the side-effect of transferring identity. Understanding this dual role enabled us to control query feature injection (Q injection) and demonstrate two applications: (1) a zero-shot motion transfer method that is 20 times more efficient than existing approaches, and (2) a training-free technique for consistent multi-shot video generation, where characters maintain identity across multiple video shots while Q injection enhances motion fidelity.
comment: (1) Project page: https://research.nvidia.com/labs/par/MotionByQueries/ (2) The methods and results in section 5, "Consistent multi-shot video generation", are based on the arXiv version 1 (v1) of this work. Here, in version 2 (v2), we extend and further analyze those findings to efficient motion transfer
♻ ☆ NeRF-Aug: Data Augmentation for Robotics with Neural Radiance Fields
Training a policy that can generalize to unknown objects is a long standing challenge within the field of robotics. The performance of a policy often drops significantly in situations where an object in the scene was not seen during training. To solve this problem, we present NeRF-Aug, a novel method that is capable of teaching a policy to interact with objects that are not present in the dataset. This approach differs from existing approaches by leveraging the speed, photorealism, and 3D consistency of a neural radiance field for augmentation. NeRF-Aug both creates more photorealistic data and runs 63% faster than existing methods. We demonstrate the effectiveness of our method on 5 tasks with 9 novel objects that are not present in the expert demonstrations. We achieve an average performance boost of 55.6% when comparing our method to the next best method. You can see video results at https://nerf-aug.github.io.
♻ ☆ Real-Time Incremental Explanations for Object Detectors in Autonomous Driving
Object detectors are widely used in safety-critical real-time applications such as autonomous driving. Explainability is especially important for safety-critical applications, and due to the variety of object detectors and their often proprietary nature, black-box explainability tools are needed. However, existing black-box explainability tools for AI models rely on multiple model calls, rendering them impractical for real-time use. In this paper, we introduce IncX, an algorithm and a tool for real-time black-box explainability for object detectors. The algorithm is based on linear transformations of saliency maps, producing sufficient explanations. We evaluate our implementation on four widely used video datasets of autonomous driving and demonstrate that IncX's explanations are comparable in quality to the state-of-the-art and are computed two orders of magnitude faster than the state-of-the-art, making them usable in real time.
♻ ☆ DepthCues: Evaluating Monocular Depth Perception in Large Vision Models CVPR 2025
Large-scale pre-trained vision models are becoming increasingly prevalent, offering expressive and generalizable visual representations that benefit various downstream tasks. Recent studies on the emergent properties of these models have revealed their high-level geometric understanding, in particular in the context of depth perception. However, it remains unclear how depth perception arises in these models without explicit depth supervision provided during pre-training. To investigate this, we examine whether the monocular depth cues, similar to those used by the human visual system, emerge in these models. We introduce a new benchmark, DepthCues, designed to evaluate depth cue understanding, and present findings across 20 diverse and representative pre-trained vision models. Our analysis shows that human-like depth cues emerge in more recent larger models. We also explore enhancing depth perception in large vision models by fine-tuning on DepthCues, and find that even without dense depth supervision, this improves depth estimation. To support further research, our benchmark and evaluation code will be made publicly available for studying depth perception in vision models.
comment: Accepted to CVPR 2025. Project page: https://danier97.github.io/depthcues/
♻ ☆ AutoLUT: LUT-Based Image Super-Resolution with Automatic Sampling and Adaptive Residual Learning CVPR2025
In recent years, the increasing popularity of Hi-DPI screens has driven a rising demand for high-resolution images. However, the limited computational power of edge devices poses a challenge in deploying complex super-resolution neural networks, highlighting the need for efficient methods. While prior works have made significant progress, they have not fully exploited pixel-level information. Moreover, their reliance on fixed sampling patterns limits both accuracy and the ability to capture fine details in low-resolution images. To address these challenges, we introduce two plug-and-play modules designed to capture and leverage pixel information effectively in Look-Up Table (LUT) based super-resolution networks. Our method introduces Automatic Sampling (AutoSample), a flexible LUT sampling approach where sampling weights are automatically learned during training to adapt to pixel variations and expand the receptive field without added inference cost. We also incorporate Adaptive Residual Learning (AdaRL) to enhance inter-layer connections, enabling detailed information flow and improving the network's ability to reconstruct fine details. Our method achieves significant performance improvements on both MuLUT and SPF-LUT while maintaining similar storage sizes. Specifically, for MuLUT, we achieve a PSNR improvement of approximately +0.20 dB improvement on average across five datasets. For SPF-LUT, with more than a 50% reduction in storage space and about a 2/3 reduction in inference time, our method still maintains performance comparable to the original. The code is available at https://github.com/SuperKenVery/AutoLUT.
comment: Accepted by CVPR2025
♻ ☆ Benchmarking Vision Language Model Unlearning via Fictitious Facial Identity Dataset
Machine unlearning has emerged as an effective strategy for forgetting specific information in the training data. However, with the increasing integration of visual data, privacy concerns in Vision Language Models (VLMs) remain underexplored. To address this, we introduce Facial Identity Unlearning Benchmark (FIUBench), a novel VLM unlearning benchmark designed to robustly evaluate the effectiveness of unlearning algorithms under the Right to be Forgotten setting. Specifically, we formulate the VLM unlearning task via constructing the Fictitious Facial Identity VQA dataset and apply a two-stage evaluation pipeline that is designed to precisely control the sources of information and their exposure levels. In terms of evaluation, since VLM supports various forms of ways to ask questions with the same semantic meaning, we also provide robust evaluation metrics including membership inference attacks and carefully designed adversarial privacy attacks to evaluate the performance of algorithms. Through the evaluation of four baseline VLM unlearning algorithms within FIUBench, we find that all methods remain limited in their unlearning performance, with significant trade-offs between model utility and forget quality. Furthermore, our findings also highlight the importance of privacy attacks for robust evaluations. We hope FIUBench will drive progress in developing more effective VLM unlearning algorithms.
♻ ☆ Depth Completion with Multiple Balanced Bases and Confidence for Dense Monocular SLAM
Dense SLAM based on monocular cameras does indeed have immense application value in the field of AR/VR, especially when it is performed on a mobile device. In this paper, we propose a novel method that integrates a light-weight depth completion network into a sparse SLAM system using a multi-basis depth representation, so that dense mapping can be performed online even on a mobile phone. Specifically, we present a specifically optimized multi-basis depth completion network, called BBC-Net, tailored to the characteristics of traditional sparse SLAM systems. BBC-Net can predict multiple balanced bases and a confidence map from a monocular image with sparse points generated by off-the-shelf keypoint-based SLAM systems. The final depth is a linear combination of predicted depth bases that can be optimized by tuning the corresponding weights. To seamlessly incorporate the weights into traditional SLAM optimization and ensure efficiency and robustness, we design a set of depth weight factors, which makes our network a versatile plug-in module, facilitating easy integration into various existing sparse SLAM systems and significantly enhancing global depth consistency through bundle adjustment. To verify the portability of our method, we integrate BBC-Net into two representative SLAM systems. The experimental results on various datasets show that the proposed method achieves better performance in monocular dense mapping than the state-of-the-art methods. We provide an online demo running on a mobile phone, which verifies the efficiency and mapping quality of the proposed method in real-world scenarios.
♻ ☆ MicroMIL: Graph-based Contextual Multiple Instance Learning for Patient Diagnosis Using Microscopy Images
Cancer diagnosis has greatly benefited from the integration of whole-slide images (WSIs) with multiple instance learning (MIL), enabling high-resolution analysis of tissue morphology. Graph-based MIL (GNN-MIL) approaches have emerged as powerful solutions for capturing spatial and relational structures in WSIs, thereby improving diagnostic accuracy. However, despite their effectiveness, WSIs require significant computational and infrastructural resources, limiting accessibility in resource-constrained settings. Microscopy imaging provides a cost-effective alternative, but applying GNN-MIL to microscopy imaging is challenging due to the absence of spatial coordinates and the high redundancy in pathologist-acquired images. To address these issues, we introduce MicroMIL, the first weakly-supervised MIL framework specifically designed for microscopy imaging. MicroMIL leverages a representative image extractor (RIE) that employs deep cluster embedding (DCE) and hard Gumbel-Softmax to dynamically reduce redundancy and select representative images. These selected images serve as graph nodes, with edges determined by cosine similarity, eliminating the need for spatial coordinates while preserving relational structure. Extensive experiments on a real-world colon cancer dataset and the BreakHis dataset demonstrate that MicroMIL achieves state-of-the-art performance, improving both diagnostic accuracy and robustness to redundancy. The code is available at https://anonymous.4open.science/r/MicroMIL-6C7C
comment: The first two authors contributed equally to this work
♻ ☆ Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion CVPR 2025
In this paper, we introduce the Selective Image Guided Network (SigNet), a novel degradation-aware framework that transforms depth completion into depth enhancement for the first time. Moving beyond direct completion using convolutional neural networks (CNNs), SigNet initially densifies sparse depth data through non-CNN densification tools to obtain coarse yet dense depth. This approach eliminates the mismatch and ambiguity caused by direct convolution over irregularly sampled sparse data. Subsequently, SigNet redefines completion as enhancement, establishing a self-supervised degradation bridge between the coarse depth and the targeted dense depth for effective RGB-D fusion. To achieve this, SigNet leverages the implicit degradation to adaptively select high-frequency components (e.g., edges) of RGB data to compensate for the coarse depth. This degradation is further integrated into a multi-modal conditional Mamba, dynamically generating the state parameters to enable efficient global high-frequency information interaction. We conduct extensive experiments on the NYUv2, DIML, SUN RGBD, and TOFDC datasets, demonstrating the state-of-the-art (SOTA) performance of SigNet.
comment: CVPR 2025
♻ ☆ PRAM: Place Recognition Anywhere Model for Efficient Visual Localization
Visual localization is a key technique to a variety of applications, e.g., autonomous driving, AR/VR, and robotics. For these real applications, both efficiency and accuracy are important especially on edge devices with limited computing resources. However, previous frameworks, e.g., absolute pose regression (APR), scene coordinate regression (SCR), and the hierarchical method (HM), have limited either accuracy or efficiency in both indoor and outdoor environments. In this paper, we propose the place recognition anywhere model (PRAM), a new framework, to perform visual localization efficiently and accurately by recognizing 3D landmarks. Specifically, PRAM first generates landmarks directly in 3D space in a self-supervised manner. Without relying on commonly used classic semantic labels, these 3D landmarks can be defined in any place in indoor and outdoor scenes with higher generalization ability. Representing the map with 3D landmarks, PRAM discards global descriptors, repetitive local descriptors, and redundant 3D points, increasing the memory efficiency significantly. Then, sparse keypoints, rather than dense pixels, are utilized as the input tokens to a transformer-based recognition module for landmark recognition, which enables PRAM to recognize hundreds of landmarks with high time and memory efficiency. At test time, sparse keypoints and predicted landmark labels are utilized for outlier removal and landmark-wise 2D-3D matching as opposed to exhaustive 2D-2D matching, which further increases the time efficiency. A comprehensive evaluation of APRs, SCRs, HMs, and PRAM on both indoor and outdoor datasets demonstrates that PRAM outperforms ARPs and SCRs in large-scale scenes with a large margin and gives competitive accuracy to HMs but reduces over 90\% memory cost and runs 2.4 times faster, leading to a better balance between efficiency and accuracy.
comment: project page: https://feixue94.github.io/pram-project/
♻ ☆ OASIS Uncovers: High-Quality T2I Models, Same Old Stereotypes ICLR 2025
Images generated by text-to-image (T2I) models often exhibit visual biases and stereotypes of concepts such as culture and profession. Existing quantitative measures of stereotypes are based on statistical parity that does not align with the sociological definition of stereotypes and, therefore, incorrectly categorizes biases as stereotypes. Instead of oversimplifying stereotypes as biases, we propose a quantitative measure of stereotypes that aligns with its sociological definition. We then propose OASIS to measure the stereotypes in a generated dataset and understand their origins within the T2I model. OASIS includes two scores to measure stereotypes from a generated image dataset: (M1) Stereotype Score to measure the distributional violation of stereotypical attributes, and (M2) WALS to measure spectral variance in the images along a stereotypical attribute. OASIS also includes two methods to understand the origins of stereotypes in T2I models: (U1) StOP to discover attributes that the T2I model internally associates with a given concept, and (U2) SPI to quantify the emergence of stereotypical attributes in the latent space of the T2I model during image generation. Despite the considerable progress in image fidelity, using OASIS, we conclude that newer T2I models such as FLUX.1 and SDv3 contain strong stereotypical predispositions about concepts and still generate images with widespread stereotypical attributes. Additionally, the quantity of stereotypes worsens for nationalities with lower Internet footprints.
comment: Accepted as a Spotlight paper at ICLR 2025
♻ ☆ ATRNet-STAR: A Large Dataset and Benchmark Towards Remote Sensing Object Recognition in the Wild
The absence of publicly available, large-scale, high-quality datasets for Synthetic Aperture Radar Automatic Target Recognition (SAR ATR) has significantly hindered the application of rapidly advancing deep learning techniques, which hold huge potential to unlock new capabilities in this field. This is primarily because collecting large volumes of diverse target samples from SAR images is prohibitively expensive, largely due to privacy concerns, the characteristics of microwave radar imagery perception, and the need for specialized expertise in data annotation. Throughout the history of SAR ATR research, there have been only a number of small datasets, mainly including targets like ships, airplanes, buildings, etc. There is only one vehicle dataset MSTAR collected in the 1990s, which has been a valuable source for SAR ATR. To fill this gap, this paper introduces a large-scale, new dataset named ATRNet-STAR with 40 different vehicle categories collected under various realistic imaging conditions and scenes. It marks a substantial advancement in dataset scale and diversity, comprising over 190,000 well-annotated samples, 10 times larger than its predecessor, the famous MSTAR. Building such a large dataset is a challenging task, and the data collection scheme will be detailed. Secondly, we illustrate the value of ATRNet-STAR via extensively evaluating the performance of 15 representative methods with 7 different experimental settings on challenging classification and detection benchmarks derived from the dataset. Finally, based on our extensive experiments, we identify valuable insights for SAR ATR and discuss potential future research directions in this field. We hope that the scale, diversity, and benchmark of ATRNet-STAR can significantly facilitate the advancement of SAR ATR.
comment: 17 pages, 14 figures; ATRNet-STAR: https://github.com/waterdisappear/ATRNet-STAR
♻ ☆ Vulnerabilities in AI-generated Image Detection: The Challenge of Adversarial Attacks
Recent advancements in image synthesis, particularly with the advent of GAN and Diffusion models, have amplified public concerns regarding the dissemination of disinformation. To address such concerns, numerous AI-generated Image (AIGI) Detectors have been proposed and achieved promising performance in identifying fake images. However, there still lacks a systematic understanding of the adversarial robustness of AIGI detectors. In this paper, we examine the vulnerability of state-of-the-art AIGI detectors against adversarial attack under white-box and black-box settings, which has been rarely investigated so far. To this end, we propose a new method to attack AIGI detectors. First, inspired by the obvious difference between real images and fake images in the frequency domain, we add perturbations under the frequency domain to push the image away from its original frequency distribution. Second, we explore the full posterior distribution of the surrogate model to further narrow this gap between heterogeneous AIGI detectors, e.g. transferring adversarial examples across CNNs and ViTs. This is achieved by introducing a novel post-train Bayesian strategy that turns a single surrogate into a Bayesian one, capable of simulating diverse victim models using one pre-trained surrogate, without the need for re-training. We name our method as Frequency-based Post-train Bayesian Attack, or FPBA. Through FPBA, we show that adversarial attack is truly a real threat to AIGI detectors, because FPBA can deliver successful black-box attacks across models, generators, defense methods, and even evade cross-generator detection, which is a crucial real-world detection scenario. The code will be shared upon acceptance.
♻ ☆ MVCTrack: Boosting 3D Point Cloud Tracking via Multimodal-Guided Virtual Cues ICRA 2025
3D single object tracking is essential in autonomous driving and robotics. Existing methods often struggle with sparse and incomplete point cloud scenarios. To address these limitations, we propose a Multimodal-guided Virtual Cues Projection (MVCP) scheme that generates virtual cues to enrich sparse point clouds. Additionally, we introduce an enhanced tracker MVCTrack based on the generated virtual cues. Specifically, the MVCP scheme seamlessly integrates RGB sensors into LiDAR-based systems, leveraging a set of 2D detections to create dense 3D virtual cues that significantly improve the sparsity of point clouds. These virtual cues can naturally integrate with existing LiDAR-based 3D trackers, yielding substantial performance gains. Extensive experiments demonstrate that our method achieves competitive performance on the NuScenes dataset.
comment: Accepted by ICRA 2025
♻ ☆ Toward Robust Non-Transferable Learning: A Survey and Benchmark
Over the past decades, researchers have primarily focused on improving the generalization abilities of models, with limited attention given to regulating such generalization. However, the ability of models to generalize to unintended data (e.g., harmful or unauthorized data) can be exploited by malicious adversaries in unforeseen ways, potentially resulting in violations of model ethics. Non-transferable learning (NTL), a task aimed at reshaping the generalization abilities of deep learning models, was proposed to address these challenges. While numerous methods have been proposed in this field, a comprehensive review of existing progress and a thorough analysis of current limitations remain lacking. In this paper, we bridge this gap by presenting the first comprehensive survey on NTL and introducing NTLBench, the first benchmark to evaluate NTL performance and robustness within a unified framework. Specifically, we first introduce the task settings, general framework, and criteria of NTL, followed by a summary of NTL approaches. Furthermore, we emphasize the often-overlooked issue of robustness against various attacks that can destroy the non-transferable mechanism established by NTL. Experiments conducted via NTLBench verify the limitations of existing NTL methods in robustness. Finally, we discuss the practical applications of NTL, along with its future directions and associated challenges.
comment: Code is available at https://github.com/tmllab/NTLBench
♻ ☆ Revisiting the Generalization Problem of Low-level Vision Models Through the Lens of Image Deraining
Generalization remains a significant challenge for low-level vision models, which often struggle with unseen degradations in real-world scenarios despite their success in controlled benchmarks. In this paper, we revisit the generalization problem in low-level vision models. Image deraining is selected as a case study due to its well-defined and easily decoupled structure, allowing for more effective observation and analysis. Through comprehensive experiments, we reveal that the generalization issue is not primarily due to limited network capacity but rather the failure of existing training strategies, which leads networks to overfit specific degradation patterns. Our findings show that guiding networks to focus on learning the underlying image content, rather than the degradation patterns, is key to improving generalization. We demonstrate that balancing the complexity of background images and degradations in the training data helps networks better fit the image distribution. Furthermore, incorporating content priors from pre-trained generative models significantly enhances generalization. Experiments on both image deraining and image denoising validate the proposed strategies. We believe the insights and solutions will inspire further research and improve the generalization of low-level vision models.
comment: arXiv admin note: substantial text overlap with arXiv:2305.15134
♻ ☆ A Simple and Generalist Approach for Panoptic Segmentation
Panoptic segmentation is an important computer vision task, where the current state-of-the-art solutions require specialized components to perform well. We propose a simple generalist framework based on a deep encoder - shallow decoder architecture with per-pixel prediction. Essentially fine-tuning a massively pretrained image model with minimal additional components. Naively this method does not yield good results. We show that this is due to imbalance during training and propose a novel method for reducing it - centroid regression in the space of spectral positional embeddings. Our method achieves panoptic quality (PQ) of 55.1 on the challenging MS-COCO dataset, state-of-the-art performance among generalist methods.
♻ ☆ Analysis of the BraTS 2023 Intracranial Meningioma Segmentation Challenge MICCAI
We describe the design and results from the BraTS 2023 Intracranial Meningioma Segmentation Challenge. The BraTS Meningioma Challenge differed from prior BraTS Glioma challenges in that it focused on meningiomas, which are typically benign extra-axial tumors with diverse radiologic and anatomical presentation and a propensity for multiplicity. Nine participating teams each developed deep-learning automated segmentation models using image data from the largest multi-institutional systematically expert annotated multilabel multi-sequence meningioma MRI dataset to date, which included 1000 training set cases, 141 validation set cases, and 283 hidden test set cases. Each case included T2, FLAIR, T1, and T1Gd brain MRI sequences with associated tumor compartment labels delineating enhancing tumor, non-enhancing tumor, and surrounding non-enhancing FLAIR hyperintensity. Participant automated segmentation models were evaluated and ranked based on a scoring system evaluating lesion-wise metrics including dice similarity coefficient (DSC) and 95% Hausdorff Distance. The top ranked team had a lesion-wise median dice similarity coefficient (DSC) of 0.976, 0.976, and 0.964 for enhancing tumor, tumor core, and whole tumor, respectively and a corresponding average DSC of 0.899, 0.904, and 0.871, respectively. These results serve as state-of-the-art benchmarks for future pre-operative meningioma automated segmentation algorithms. Additionally, we found that 1286 of 1424 cases (90.3%) had at least 1 compartment voxel abutting the edge of the skull-stripped image edge, which requires further investigation into optimal pre-processing face anonymization steps.
comment: Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA) https://melba-journal.org/2025:003 22 pages, 6 tables, 12 figures, MICCAI, MELBA
♻ ☆ RS-vHeat: Heat Conduction Guided Efficient Remote Sensing Foundation Model
Remote sensing foundation models largely break away from the traditional paradigm of designing task-specific models, offering greater scalability across multiple tasks. However, they face challenges such as low computational efficiency and limited interpretability, especially when dealing with large-scale remote sensing images. To overcome these, we draw inspiration from heat conduction, a physical process modeling local heat diffusion. Building on this idea, we are the first to explore the potential of using the parallel computing model of heat conduction to simulate the local region correlations in high-resolution remote sensing images, and introduce RS-vHeat, an efficient multi-modal remote sensing foundation model. Specifically, RS-vHeat 1) applies the Heat Conduction Operator (HCO) with a complexity of $O(N^{1.5})$ and a global receptive field, reducing computational overhead while capturing remote sensing object structure information to guide heat diffusion; 2) learns the frequency distribution representations of various scenes through a self-supervised strategy based on frequency domain hierarchical masking and multi-domain reconstruction; 3) significantly improves efficiency and performance over state-of-the-art techniques across 4 tasks and 10 datasets. Compared to attention-based remote sensing foundation models, we reduce memory usage by 84\%, FLOPs by 24\% and improves throughput by 2.7 times. The code will be made publicly available.
comment: 19 pages, 8 figures and 10 tables
♻ ☆ Language-guided Medical Image Segmentation with Target-informed Multi-level Contrastive Alignments
Medical image segmentation is crucial in modern medical image analysis, which can aid into diagnosis of various disease conditions. Recently, language-guided segmentation methods have shown promising results in automating image segmentation where text reports are incorporated as guidance. These text reports, containing image impressions and insights given by clinicians, provides auxiliary guidance. However, these methods neglect the inherent pattern gaps between the two distinct modalities, which leads to sub-optimal image-text feature fusion without proper cross-modality feature alignments. Contrastive alignments are widely used to associate image-text semantics in representation learning; however, it has not been exploited to bridge the pattern gaps in language-guided segmentation that relies on subtle low level image details to represent diseases. Existing contrastive alignment methods typically algin high-level global image semantics without involving low-level, localized target information, and therefore fails to explore fine-grained text guidance for language-guided segmentation. In this study, we propose a language-guided segmentation network with Target-informed Multi-level Contrastive Alignments (TMCA). TMCA enables target-informed cross-modality alignments and fine-grained text guidance to bridge the pattern gaps in language-guided segmentation. Specifically, we introduce: 1) a target-sensitive semantic distance module that enables granular image-text alignment modelling, and 2) a multi-level alignment strategy that directs text guidance on low-level image features. In addition, a language-guided target enhancement module is proposed to leverage the aligned text to redirect attention to focus on critical localized image features. Extensive experiments on 4 image-text datasets, involving 3 medical imaging modalities, demonstrated that our TMCA achieved superior performances.
♻ ☆ A Survey on 3D Gaussian Splatting
3D Gaussian splatting (GS) has emerged as a transformative technique in explicit radiance field and computer graphics. This innovative approach, characterized by the use of millions of learnable 3D Gaussians, represents a significant departure from mainstream neural radiance field approaches, which predominantly use implicit, coordinate-based models to map spatial coordinates to pixel values. 3D GS, with its explicit scene representation and differentiable rendering algorithm, not only promises real-time rendering capability but also introduces unprecedented levels of editability. This positions 3D GS as a potential game-changer for the next generation of 3D reconstruction and representation. In the present paper, we provide the first systematic overview of the recent developments and critical contributions in the domain of 3D GS. We begin with a detailed exploration of the underlying principles and the driving forces behind the emergence of 3D GS, laying the groundwork for understanding its significance. A focal point of our discussion is the practical applicability of 3D GS. By enabling unprecedented rendering speed, 3D GS opens up a plethora of applications, ranging from virtual reality to interactive media and beyond. This is complemented by a comparative analysis of leading 3D GS models, evaluated across various benchmark tasks to highlight their performance and practical utility. The survey concludes by identifying current challenges and suggesting potential avenues for future research. Through this survey, we aim to provide a valuable resource for both newcomers and seasoned researchers, fostering further exploration and advancement in explicit radiance field.
comment: Ongoing project. Paper list: https://github.com/guikunchen/Awesome3DGS ; Benchmark: https://github.com/guikunchen/3DGS-Benchmarks
♻ ☆ Gaussians-to-Life: Text-Driven Animation of 3D Gaussian Splatting Scenes 3DV 2025
State-of-the-art novel view synthesis methods achieve impressive results for multi-view captures of static 3D scenes. However, the reconstructed scenes still lack "liveliness," a key component for creating engaging 3D experiences. Recently, novel video diffusion models generate realistic videos with complex motion and enable animations of 2D images, however they cannot naively be used to animate 3D scenes as they lack multi-view consistency. To breathe life into the static world, we propose Gaussians2Life, a method for animating parts of high-quality 3D scenes in a Gaussian Splatting representation. Our key idea is to leverage powerful video diffusion models as the generative component of our model and to combine these with a robust technique to lift 2D videos into meaningful 3D motion. We find that, in contrast to prior work, this enables realistic animations of complex, pre-existing 3D scenes and further enables the animation of a large variety of object classes, while related work is mostly focused on prior-based character animation, or single 3D objects. Our model enables the creation of consistent, immersive 3D experiences for arbitrary scenes.
comment: Project website at https://wimmerth.github.io/gaussians2life.html. Accepted to 3DV 2025
♻ ☆ Sustainable transparency in Recommender Systems: Bayesian Ranking of Images for Explainability
Recommender Systems have become crucial in the modern world, commonly guiding users towards relevant content or products, and having a large influence over the decisions of users and citizens. However, ensuring transparency and user trust in these systems remains a challenge; personalized explanations have emerged as a solution, offering justifications for recommendations. Among the existing approaches for generating personalized explanations, using existing visual content created by users is a promising option to maximize transparency and user trust. State-of-the-art models that follow this approach, despite leveraging highly optimized architectures, employ surrogate learning tasks that do not efficiently model the objective of ranking images as explanations for a given recommendation; this leads to a suboptimal training process with high computational costs that may not be reduced without affecting model performance. This work presents BRIE, a novel model where we leverage Bayesian Pairwise Ranking to enhance the training process, allowing us to consistently outperform state-of-the-art models in six real-world datasets while reducing its model size by up to 64 times and its CO2 emissions by up to 75% in training and inference.
♻ ☆ Bridging Text and Vision: A Multi-View Text-Vision Registration Approach for Cross-Modal Place Recognition
Mobile robots necessitate advanced natural language understanding capabilities to accurately identify locations and perform tasks such as package delivery. However, traditional visual place recognition (VPR) methods rely solely on single-view visual information and cannot interpret human language descriptions. To overcome this challenge, we bridge text and vision by proposing a multiview (360{\deg} views of the surroundings) text-vision registration approach called Text4VPR for place recognition task, which is the first method that exclusively utilizes textual descriptions to match a database of images. Text4VPR employs the frozen T5 language model to extract global textual embeddings. Additionally, it utilizes the Sinkhorn algorithm with temperature coefficient to assign local tokens to their respective clusters, thereby aggregating visual descriptors from images. During the training stage, Text4VPR emphasizes the alignment between individual text-image pairs for precise textual description. In the inference stage, Text4VPR uses the Cascaded Cross-Attention Cosine Alignment (CCCA) to address the internal mismatch between text and image groups. Subsequently, Text4VPR performs precisely place match based on the descriptions of text-image groups. On Street360Loc, the first text to image VPR dataset we created, Text4VPR builds a robust baseline, achieving a leading top-1 accuracy of 57% and a leading top-10 accuracy of 92% within a 5-meter radius on the test set, which indicates that localization from textual descriptions to images is not only feasible but also holds significant potential for further advancement, as shown in Figure 1.
comment: 8 pages, 4 figures, conference
♻ ☆ A Hybrid SNN-ANN Network for Event-based Object Detection with Spatial and Temporal AttentionEfficient Event-Based Object Detection: A Hybrid Neural Network with Spatial and Temporal Attention
Event cameras offer high temporal resolution and dynamic range with minimal motion blur, making them promising for robust object detection. While Spiking Neural Networks (SNNs) on neuromorphic hardware are often considered for energy efficient and low latency event-based data processing, they often fall short of Artificial Neural Networks (ANNs) in accuracy and flexibility. Here, we introduce Attention-based Hybrid SNN-ANN backbones for event-based object detection to leverage the strengths of both SNN and ANN architectures. A novel Attention-based SNN-ANN bridge module captures sparse spatial and temporal relations from the SNN layer and converts them into dense feature maps for the ANN part of the backbone. Additionally, we present a variant that integrates DWConvLSTMs to the ANN blocks to capture slower dynamics. This multi-timescale network combines fast SNN processing for short timesteps with long-term dense RNN processing, effectively capturing both fast and slow dynamics. Experimental results demonstrate that our proposed method surpasses SNN-based approaches by significant margins, with results comparable to existing ANN and RNN-based methods. Unlike ANN-only networks, the hybrid setup allows us to implement the SNN blocks on digital neuromorphic hardware to investigate the feasibility of our approach. Extensive ablation studies and implementation on neuromorphic hardware confirm the effectiveness of our proposed modules and architectural choices. Our hybrid SNN-ANN architectures pave the way for ANN-like performance at a drastically reduced parameter, latency, and power budget.
♻ ☆ General Detection-based Text Line Recognition
We introduce a general detection-based approach to text line recognition, be it printed (OCR) or handwritten (HTR), with Latin, Chinese, or ciphered characters. Detection-based approaches have until now been largely discarded for HTR because reading characters separately is often challenging, and character-level annotation is difficult and expensive. We overcome these challenges thanks to three main insights: (i) synthetic pre-training with sufficiently diverse data enables learning reasonable character localization for any script; (ii) modern transformer-based detectors can jointly detect a large number of instances, and, if trained with an adequate masking strategy, leverage consistency between the different detections; (iii) once a pre-trained detection model with approximate character localization is available, it is possible to fine-tune it with line-level annotation on real data, even with a different alphabet. Our approach, dubbed DTLR, builds on a completely different paradigm than state-of-the-art HTR methods, which rely on autoregressive decoding, predicting character values one by one, while we treat a complete line in parallel. Remarkably, we demonstrate good performance on a large range of scripts, usually tackled with specialized approaches. In particular, we improve state-of-the-art performances for Chinese script recognition on the CASIA v2 dataset, and for cipher recognition on the Borg and Copiale datasets. Our code and models are available at https://github.com/raphael-baena/DTLR.
♻ ☆ METDrive: Multi-modal End-to-end Autonomous Driving with Temporal Guidance ICRA
Multi-modal end-to-end autonomous driving has shown promising advancements in recent work. By embedding more modalities into end-to-end networks, the system's understanding of both static and dynamic aspects of the driving environment is enhanced, thereby improving the safety of autonomous driving. In this paper, we introduce METDrive, an end-to-end system that leverages temporal guidance from the embedded time series features of ego states, including rotation angles, steering, throttle signals, and waypoint vectors. The geometric features derived from perception sensor data and the time series features of ego state data jointly guide the waypoint prediction with the proposed temporal guidance loss function. We evaluated METDrive on the CARLA leaderboard benchmarks, achieving a driving score of 70%, a route completion score of 94%, and an infraction score of 0.78.
comment: Accepted by ICRA
♻ ☆ A Simple yet Effective Test-Time Adaptation for Zero-Shot Monocular Metric Depth Estimation
The recent development of foundation models for monocular depth estimation such as Depth Anything paved the way to zero-shot monocular depth estimation. Since it returns an affine-invariant disparity map, the favored technique to recover the metric depth consists in fine-tuning the model. However, this stage is not straightforward, it can be costly and time-consuming because of the training and the creation of the dataset. The latter must contain images captured by the camera that will be used at test time and the corresponding ground truth. Moreover, the fine-tuning may also degrade the generalizing capacity of the original model. Instead, we propose in this paper a new method to rescale Depth Anything predictions using 3D points provided by sensors or techniques such as low-resolution LiDAR or structure-from-motion with poses given by an IMU. This approach avoids fine-tuning and preserves the generalizing power of the original depth estimation model while being robust to the noise of the sparse depth or of the depth model. Our experiments highlight enhancements relative to zero-shot monocular metric depth estimation methods, competitive results compared to fine-tuned approaches and a better robustness than depth completion approaches. Code available at https://gitlab.ensta.fr/ssh/monocular-depth-rescaling.
♻ ☆ SimpleDepthPose: Fast and Reliable Human Pose Estimation with RGBD-Images
In the rapidly advancing domain of computer vision, accurately estimating the poses of multiple individuals from various viewpoints remains a significant challenge, especially when reliability is a key requirement. This paper introduces a novel algorithm that excels in multi-view, multi-person pose estimation by incorporating depth information. An extensive evaluation demonstrates that the proposed algorithm not only generalizes well to unseen datasets, and shows a fast runtime performance, but also is adaptable to different keypoints. To support further research, all of the work is publicly accessible.
♻ ☆ Needle In A Video Haystack: A Scalable Synthetic Evaluator for Video MLLMs ICLR 2025
Video understanding is a crucial next step for multimodal large language models (MLLMs). Various benchmarks are introduced for better evaluating the MLLMs. Nevertheless, current video benchmarks are still inefficient for evaluating video models during iterative development due to the high cost of constructing datasets and the difficulty in isolating specific skills. In this paper, we propose VideoNIAH (Video Needle In A Haystack), a benchmark construction framework through synthetic video generation. VideoNIAH decouples video content from their query-responses by inserting unrelated visual 'needles' into original videos. The framework automates the generation of query-response pairs using predefined rules, minimizing manual labor. The queries focus on specific aspects of video understanding, enabling more skill-specific evaluations. The separation between video content and the queries also allow for increased video variety and evaluations across different lengths. Utilizing VideoNIAH, we compile a video benchmark VNBench, which includes tasks such as retrieval, ordering, and counting to evaluate three key aspects of video understanding: temporal perception, chronological ordering, and spatio-temporal coherence. We conduct a comprehensive evaluation of both proprietary and open-source models, uncovering significant differences in their video understanding capabilities across various tasks. Additionally, we perform an in-depth analysis of the test results and model configurations. Based on these findings, we provide some advice for improving video MLLM training, offering valuable insights to guide future research and model development. The code and data are available at https://github.com/joez17/VideoNIAH.
comment: ICLR 2025
♻ ☆ Large Language Models are Strong Audio-Visual Speech Recognition Learners ICASSP 2025
Multimodal large language models (MLLMs) have recently become a focal point of research due to their formidable multimodal understanding capabilities. For example, in the audio and speech domains, an LLM can be equipped with (automatic) speech recognition (ASR) abilities by just concatenating the audio tokens, computed with an audio encoder, and the text tokens to achieve state-of-the-art results. On the contrary, tasks like visual and audio-visual speech recognition (VSR/AVSR), which also exploit noise-invariant lip movement information, have received little or no attention. To bridge this gap, we propose Llama-AVSR, a new MLLM with strong audio-visual speech recognition capabilities. It leverages pre-trained audio and video encoders to produce modality-specific tokens which, together with the text tokens, are processed by a pre-trained LLM (e.g., Llama3.1-8B) to yield the resulting response in an auto-regressive fashion. Llama-AVSR requires a small number of trainable parameters as only modality-specific projectors and LoRA modules are trained whereas the multi-modal encoders and LLM are kept frozen. We evaluate our proposed approach on LRS3, the largest public AVSR benchmark, and we achieve new state-of-the-art results for the tasks of ASR and AVSR with a WER of 0.79% and 0.77%, respectively. To bolster our results, we investigate the key factors that underpin the effectiveness of Llama-AVSR: the choice of the pre-trained encoders and LLM, the efficient integration of LoRA modules, and the optimal performance-efficiency trade-off obtained via modality-aware compression rates.
comment: Accepted for publication at ICASSP 2025. The code and checkpoints are available here: https://github.com/umbertocappellazzo/Llama-AVSR
♻ ☆ Nexus-O: An Omni-Perceptive And -Interactive Model for Language, Audio, And Vision
Human beings perceive the real world through a spectrum of sensory modalities, encompassing auditory, visual, and linguistic faculties. The journey towards achieving Artificial General Intelligence (AGI) necessitates the development of models that can emulate these multifaceted perceptual capabilities and comprehensively understand these diversified data. To this end, we introduce \textbf{Nexus-O}, an industry-level \textbf{omni-perceptive and -interactive} model capable of efficiently processing Audio, Image, Video, and Text data in any combination and output audio/text in an end-to-end way. We systematically investigate Nexus-O by addressing three key research questions: First, how can models be efficiently designed and trained to achieve tri-modal alignment, understanding and reasoning capabilities across multiple modalities? Second, what approaches can be implemented to evaluate tri-modal model robustness, ensuring reliable performance and applicability in real-world scenarios? Third, what strategies can be employed to curate and obtain high-quality, real-life scenario speech datasets? For the first question, we design and pre-train Nexus-O based on the vision-language model, rather than the language model. By pre-training the model over high-quality synthetic audio data, our model is capable of tri-modal perception and interaction. For the second question, we introduce a new audio testbed, Nexus-O-audio, comprising diverse Automatic Speech Recognition (ASR) samples, spanning various real-world scenarios, such as corporate meetings and live stream. For the third question, we design the speech data synthesis pipeline to obtain high-quality speech training datasets, covering various real-world scenarios. Comprehensive experimentation and an in-depth analysis of tri-modal alignment over latent space demonstrate the advantages of our model on downstream tasks.
♻ ☆ PseudoTouch: Efficiently Imaging the Surface Feel of Objects for Robotic Manipulation ICRA 2025
Tactile sensing is vital for human dexterous manipulation, however, it has not been widely used in robotics. Compact, low-cost sensing platforms can facilitate a change, but unlike their popular optical counterparts, they are difficult to deploy in high-fidelity tasks due to their low signal dimensionality and lack of a simulation model. To overcome these challenges, we introduce PseudoTouch which links high-dimensional structural information to low-dimensional sensor signals. It does so by learning a low-dimensional visual-tactile embedding, wherein we encode a depth patch from which we decode the tactile signal. We collect and train PseudoTouch on a dataset comprising aligned tactile and visual data pairs obtained through random touching of eight basic geometric shapes. We demonstrate the utility of our trained PseudoTouch model in two downstream tasks: object recognition and grasp stability prediction. In the object recognition task, we evaluate the learned embedding's performance on a set of five basic geometric shapes and five household objects. Using PseudoTouch, we achieve an object recognition accuracy 84% after just ten touches, surpassing a proprioception baseline. For the grasp stability task, we use ACRONYM labels to train and evaluate a grasp success predictor using PseudoTouch's predictions derived from virtual depth information. Our approach yields a 32% absolute improvement in accuracy compared to the baseline relying on partial point cloud data. We make the data, code, and trained models publicly available at https://pseudotouch.cs.uni-freiburg.de.
comment: 7 pages, 5 figures, 2 tables, accepted at ICRA 2025
♻ ☆ DreamForge: Motion-Aware Autoregressive Video Generation for Multi-View Driving Scenes
Recent advances in diffusion models have improved controllable streetscape generation and supported downstream perception and planning tasks. However, challenges remain in accurately modeling driving scenes and generating long videos. To alleviate these issues, we propose DreamForge, an advanced diffusion-based autoregressive video generation model tailored for 3D-controllable long-term generation. To enhance the lane and foreground generation, we introduce perspective guidance and integrate object-wise position encoding to incorporate local 3D correlation and improve foreground object modeling. We also propose motion-aware temporal attention to capture motion cues and appearance changes in videos. By leveraging motion frames and an autoregressive generation paradigm,we can autoregressively generate long videos (over 200 frames) using a model trained in short sequences, achieving superior quality compared to the baseline in 16-frame video evaluations. Finally, we integrate our method with the realistic simulator DriveArena to provide more reliable open-loop and closed-loop evaluations for vision-based driving agents. Project Page: https://pjlab-adg.github.io/DriveArena/dreamforge.
comment: 15 figures, 9 tables
♻ ☆ Modification Takes Courage: Seamless Image Stitching via Reference-Driven Inpainting
Current image stitching methods often produce noticeable seams in challenging scenarios such as uneven hue and large parallax. To tackle this problem, we propose the Reference-Driven Inpainting Stitcher (RDIStitcher), which reformulates the image fusion and rectangling as a reference-based inpainting model, incorporating a larger modification fusion area and stronger modification intensity than previous methods. Furthermore, we introduce a self-supervised model training method, which enables the implementation of RDIStitcher without requiring labeled data by fine-tuning a Text-to-Image (T2I) diffusion model. Recognizing difficulties in assessing the quality of stitched images, we present the Multimodal Large Language Models (MLLMs)-based metrics, offering a new perspective on evaluating stitched image quality. Compared to the state-of-the-art (SOTA) method, extensive experiments demonstrate that our method significantly enhances content coherence and seamless transitions in the stitched images. Especially in the zero-shot experiments, our method exhibits strong generalization capabilities. Code: https://github.com/yayoyo66/RDIStitcher
comment: 18 pages, 10 figures
♻ ☆ VidHal: Benchmarking Temporal Hallucinations in Vision LLMs
Vision Large Language Models (VLLMs) are widely acknowledged to be prone to hallucinations. Existing research addressing this problem has primarily been confined to image inputs, with limited exploration of video-based hallucinations. Furthermore, current evaluation methods fail to capture nuanced errors in generated responses, which are often exacerbated by the rich spatiotemporal dynamics of videos. To address this, we introduce VidHal, a benchmark specially designed to evaluate video-based hallucinations in VLLMs. VidHal is constructed by bootstrapping video instances across a wide range of common temporal aspects. A defining feature of our benchmark lies in the careful creation of captions which represent varying levels of hallucination associated with each video. To enable fine-grained evaluation, we propose a novel caption ordering task requiring VLLMs to rank captions by hallucinatory extent. We conduct extensive experiments on VidHal and comprehensively evaluate a broad selection of models. Our results uncover significant limitations in existing VLLMs regarding hallucination generation. Through our benchmark, we aim to inspire further research on 1) holistic understanding of VLLM capabilities, particularly regarding hallucination, and 2) extensive development of advanced VLLMs to alleviate this problem.
comment: 9 pages, 10 figures. Code available at https://github.com/Lookuz/VidHal
♻ ☆ Multi-Knowledge-oriented Nighttime Haze Imaging Enhancer for Vision-driven Intelligent Systems
Salient object detection (SOD) plays a critical role in vision-driven measurement systems (VMS), facilitating the detection and segmentation of key visual elements in an image. However, adverse imaging conditions such as haze during the day, low light, and haze at night severely degrade image quality, and complicating the SOD process. To address these challenges, we propose a multi-task-oriented nighttime haze imaging enhancer (MToIE), which integrates three tasks: daytime dehazing, low-light enhancement, and nighttime dehazing. The MToIE incorporates two key innovative components: First, the network employs a task-oriented node learning mechanism to handle three specific degradation types: day-time haze, low light, and night-time haze conditions, with an embedded self-attention module enhancing its performance in nighttime imaging. In addition, multi-receptive field enhancement module that efficiently extracts multi-scale features through three parallel depthwise separable convolution branches with different dilation rates, capturing comprehensive spatial information with minimal computational overhead. To ensure optimal image reconstruction quality and visual characteristics, we suggest a hybrid loss function. Extensive experiments on different types of weather/imaging conditions illustrate that MToIE surpasses existing methods, significantly enhancing the accuracy and reliability of vision systems across diverse imaging scenarios. The code is available at https://github.com/Ai-Chen-Lab/MKoIE.
♻ ☆ Generalized moduli of continuity under irregular or random deformations via multiscale analysis
Motivated by the problem of robustness to deformations of the input for deep convolutional neural networks, we identify signal classes which are inherently stable to irregular deformations induced by distortion fields $\tau\in L^\infty(\mathbb{R}^d;\mathbb{R}^d)$, to be characterized in terms of a generalized modulus of continuity associated with the deformation operator. Resorting to ideas of harmonic and multiscale analysis, we prove that for signals in multiresolution approximation spaces $U_s$ at scale $s$, stability in $L^2$ holds in the regime $\|\tau\|_{L^\infty}/s\ll 1$ - essentially as an effect of the uncertainty principle. Instability occurs when $\|\tau\|_{L^\infty}/s\gg 1$, and we provide a sharp upper bound for the asymptotic growth rate. The stability results are then extended to signals in the Besov space $B^{d/2}_{2,1}$ tailored to the given multiresolution approximation. We also consider the case of more general time-frequency deformations. Finally, we provide stochastic versions of the aforementioned results, namely we study the issue of stability in mean when $\tau(x)$ is modeled as a random field (not bounded, in general) with identically distributed variables $|\tau(x)|$, $x\in\mathbb{R}^d$.
comment: 25 pages
♻ ☆ DLF: Extreme Image Compression with Dual-generative Latent Fusion
Recent studies in extreme image compression have achieved remarkable performance by compressing the tokens from generative tokenizers. However, these methods often prioritize clustering common semantics within the dataset, while overlooking the diverse details of individual objects. Consequently, this results in suboptimal reconstruction fidelity, especially at low bitrates. To address this issue, we introduce a Dual-generative Latent Fusion (DLF) paradigm. DLF decomposes the latent into semantic and detail elements, compressing them through two distinct branches. The semantic branch clusters high-level information into compact tokens, while the detail branch encodes perceptually critical details to enhance the overall fidelity. Additionally, we propose a cross-branch interactive design to reduce redundancy between the two branches, thereby minimizing the overall bit cost. Experimental results demonstrate the impressive reconstruction quality of DLF even below 0.01 bits per pixel (bpp). On the CLIC2020 test set, our method achieves bitrate savings of up to 27.93% on LPIPS and 53.55% on DISTS compared to MS-ILLM. Furthermore, DLF surpasses recent diffusion-based codecs in visual fidelity while maintaining a comparable level of generative realism. Code will be available later.
♻ ☆ RURANET++: An Unsupervised Learning Method for Diabetic Macular Edema Based on SCSE Attention Mechanisms and Dynamic Multi-Projection Head Clustering MICCAI 2025
Diabetic Macular Edema (DME), a prevalent complication among diabetic patients, constitutes a major cause of visual impairment and blindness. Although deep learning has achieved remarkable progress in medical image analysis, traditional DME diagnosis still relies on extensive annotated data and subjective ophthalmologist assessments, limiting practical applications. To address this, we present RURANET++, an unsupervised learning-based automated DME diagnostic system. This framework incorporates an optimized U-Net architecture with embedded Spatial and Channel Squeeze & Excitation (SCSE) attention mechanisms to enhance lesion feature extraction. During feature processing, a pre-trained GoogLeNet model extracts deep features from retinal images, followed by PCA-based dimensionality reduction to 50 dimensions for computational efficiency. Notably, we introduce a novel clustering algorithm employing multi-projection heads to explicitly control cluster diversity while dynamically adjusting similarity thresholds, thereby optimizing intra-class consistency and inter-class discrimination. Experimental results demonstrate superior performance across multiple metrics, achieving maximum accuracy (0.8411), precision (0.8593), recall (0.8411), and F1-score (0.8390), with exceptional clustering quality. This work provides an efficient unsupervised solution for DME diagnosis with significant clinical implications.
comment: 10 pages, 2 figures, 5 tables, submitted to The 28th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2025)
♻ ☆ CLIP meets DINO for Tuning Zero-Shot Classifier using Unlabeled Image Collections
In the era of foundation models, CLIP has emerged as a powerful tool for aligning text & visual modalities into a common embedding space. However, the alignment objective used to train CLIP often results in subpar visual features for fine-grained tasks. In contrast, SSL-pretrained models like DINO excel at extracting rich visual features due to their specialized training paradigm. Yet, these SSL models require an additional supervised linear probing step, which relies on fully labeled data which is often expensive and difficult to obtain at scale. In this paper, we propose a label-free prompt-tuning method that leverages the rich visual features of self-supervised learning models (DINO) and the broad textual knowledge of large language models (LLMs) to largely enhance CLIP-based image classification performance using unlabeled images. Our approach unfolds in three key steps: (1) We generate robust textual feature embeddings that more accurately represent object classes by leveraging class-specific descriptions from LLMs, enabling more effective zero-shot classification compared to CLIP's default name-specific prompts. (2) These textual embeddings are then used to produce pseudo-labels to train an alignment module that integrates the complementary strengths of LLM description-based textual embeddings & DINO's visual features. (3) Finally, we prompt-tune CLIP's vision encoder through DINO-assisted supervision using the trained alignment module. This three-step process allows us to harness the best of visual & textual foundation models, resulting in a powerful and efficient approach that surpasses state-of-the-art label-free classification methods. Notably, our framework, NoLA (No Labels Attached), achieves an average absolute gain of 3.6% over the state-of-the-art LaFTer across 11 diverse image classification datasets. Our code & models can be found at https://github.com/fazliimam/NoLA.
♻ ☆ Towards Student Actions in Classroom Scenes: New Dataset and Baseline
Analyzing student actions is an important and challenging task in educational research. Existing efforts have been hampered by the lack of accessible datasets to capture the nuanced action dynamics in classrooms. In this paper, we present a new multi-label Student Action Video (SAV) dataset, specifically designed for action detection in classroom settings. The SAV dataset consists of 4,324 carefully trimmed video clips from 758 different classrooms, annotated with 15 distinct student actions. Compared to existing action detection datasets, the SAV dataset stands out by providing a wide range of real classroom scenarios, high-quality video data, and unique challenges, including subtle movement differences, dense object engagement, significant scale differences, varied shooting angles, and visual occlusion. These complexities introduce new opportunities and challenges to advance action detection methods. To benchmark this, we propose a novel baseline method based on a visual transformer, designed to enhance attention to key local details within small and dense object regions. Our method demonstrates excellent performance with a mean Average Precision (mAP) of 67.9% and 27.4% on the SAV and AVA datasets, respectively. This paper not only provides the dataset but also calls for further research into AI-driven educational tools that may transform teaching methodologies and learning outcomes. The code and dataset are released at https://github.com/Ritatanz/SAV.
♻ ☆ LRSAA: Large-scale Remote Sensing Image Target Recognition and Automatic Annotation
This paper presents a method for object recognition and automatic labeling in large-area remote sensing images called LRSAA. The method integrates YOLOv11 and MobileNetV3-SSD object detection algorithms through ensemble learning to enhance model performance. Furthermore, it employs Poisson disk sampling segmentation techniques and the EIOU metric to optimize the training and inference processes of segmented images, followed by the integration of results. This approach not only reduces the demand for computational resources but also achieves a good balance between accuracy and speed. The source code for this project has been made publicly available on https://github.com/anaerovane/LRSAA.
comment: arXiv admin note: text overlap with arXiv:2411.07802
♻ ☆ Raccoon: Multi-stage Diffusion Training with Coarse-to-Fine Curating Videos
Text-to-video generation has demonstrated promising progress with the advent of diffusion models, yet existing approaches are limited by dataset quality and computational resources. To address these limitations, this paper presents a comprehensive approach that advances both data curation and model design. We introduce CFC-VIDS-1M, a high-quality video dataset constructed through a systematic coarse-to-fine curation pipeline. The pipeline first evaluates video quality across multiple dimensions, followed by a fine-grained stage that leverages vision-language models to enhance text-video alignment and semantic richness. Building upon the curated dataset's emphasis on visual quality and temporal coherence, we develop RACCOON, a transformer-based architecture with decoupled spatial-temporal attention mechanisms. The model is trained through a progressive four-stage strategy designed to efficiently handle the complexities of video generation. Extensive experiments demonstrate that our integrated approach of high-quality data curation and efficient training strategy generates visually appealing and temporally coherent videos while maintaining computational efficiency. We will release our dataset, code, and models.
♻ ☆ CAT-3DGS: A Context-Adaptive Triplane Approach to Rate-Distortion-Optimized 3DGS Compression ICLR
3D Gaussian Splatting (3DGS) has recently emerged as a promising 3D representation. Much research has been focused on reducing its storage requirements and memory footprint. However, the needs to compress and transmit the 3DGS representation to the remote side are overlooked. This new application calls for rate-distortion-optimized 3DGS compression. How to quantize and entropy encode sparse Gaussian primitives in the 3D space remains largely unexplored. Few early attempts resort to the hyperprior framework from learned image compression. But, they fail to utilize fully the inter and intra correlation inherent in Gaussian primitives. Built on ScaffoldGS, this work, termed CAT-3DGS, introduces a context-adaptive triplane approach to their rate-distortion-optimized coding. It features multi-scale triplanes, oriented according to the principal axes of Gaussian primitives in the 3D space, to capture their inter correlation (i.e. spatial correlation) for spatial autoregressive coding in the projected 2D planes. With these triplanes serving as the hyperprior, we further perform channel-wise autoregressive coding to leverage the intra correlation within each individual Gaussian primitive. Our CAT-3DGS incorporates a view frequency-aware masking mechanism. It actively skips from coding those Gaussian primitives that potentially have little impact on the rendering quality. When trained end-to-end to strike a good rate-distortion trade-off, our CAT-3DGS achieves the state-of-the-art compression performance on the commonly used real-world datasets.
comment: Accepted for Publication in International Conference on Learning Representations (ICLR)
♻ ☆ NavRAG: Generating User Demand Instructions for Embodied Navigation through Retrieval-Augmented LLM
Vision-and-Language Navigation (VLN) is an essential skill for embodied agents, allowing them to navigate in 3D environments following natural language instructions. High-performance navigation models require a large amount of training data, the high cost of manually annotating data has seriously hindered this field. Therefore, some previous methods translate trajectory videos into step-by-step instructions for expanding data, but such instructions do not match well with users' communication styles that briefly describe destinations or state specific needs. Moreover, local navigation trajectories overlook global context and high-level task planning. To address these issues, we propose NavRAG, a retrieval-augmented generation (RAG) framework that generates user demand instructions for VLN. NavRAG leverages LLM to build a hierarchical scene description tree for 3D scene understanding from global layout to local details, then simulates various user roles with specific demands to retrieve from the scene tree, generating diverse instructions with LLM. We annotate over 2 million navigation instructions across 861 scenes and evaluate the data quality and navigation performance of trained models.
♻ ☆ Generative Densification: Learning to Densify Gaussians for High-Fidelity Generalizable 3D Reconstruction
Generalized feed-forward Gaussian models have achieved significant progress in sparse-view 3D reconstruction by leveraging prior knowledge from large multi-view datasets. However, these models often struggle to represent high-frequency details due to the limited number of Gaussians. While the densification strategy used in per-scene 3D Gaussian splatting (3D-GS) optimization can be adapted to the feed-forward models, it may not be ideally suited for generalized scenarios. In this paper, we propose Generative Densification, an efficient and generalizable method to densify Gaussians generated by feed-forward models. Unlike the 3D-GS densification strategy, which iteratively splits and clones raw Gaussian parameters, our method up-samples feature representations from the feed-forward models and generates their corresponding fine Gaussians in a single forward pass, leveraging the embedded prior knowledge for enhanced generalization. Experimental results on both object-level and scene-level reconstruction tasks demonstrate that our method outperforms state-of-the-art approaches with comparable or smaller model sizes, achieving notable improvements in representing fine details.
comment: Project page: https://stnamjef.github.io/GenerativeDensification/
♻ ☆ Neighboring Slice Noise2Noise: Self-Supervised Medical Image Denoising from Single Noisy Image Volume
In the last few years, with the rapid development of deep learning technologies, supervised methods based on convolutional neural networks have greatly enhanced the performance of medical image denoising. However, these methods require large quantities of noisy-clean image pairs for training, which greatly limits their practicality. Although some researchers have attempted to train denoising networks using only single noisy images, existing self-supervised methods, including blind-spot-based and data-splitting-based methods, heavily rely on the assumption that noise is pixel-wise independent. However, this assumption often does not hold in real-world medical images. Therefore, in the field of medical imaging, there remains a lack of simple and practical denoising methods that can achieve high-quality denoising performance using only single noisy images. In this paper, we propose a novel self-supervised medical image denoising method, Neighboring Slice Noise2Noise (NS-N2N). The proposed method utilizes neighboring slices within a single noisy image volume to construct weighted training data, and then trains the denoising network using a self-supervised scheme with regional consistency loss and inter-slice continuity loss. NS-N2N only requires a single noisy image volume obtained from one medical imaging procedure to achieve high-quality denoising of the image volume itself. Extensive experiments demonstrate that the proposed method outperforms state-of-the-art self-supervised denoising methods in both denoising performance and processing efficiency. Furthermore, since NS-N2N operates solely in the image domain, it is free from device-specific issues such as reconstruction geometry, making it easier to apply in various clinical practices.
♻ ☆ Meta Curvature-Aware Minimization for Domain Generalization
Domain generalization (DG) aims to enhance the ability of models trained on source domains to generalize effectively to unseen domains. Recently, Sharpness-Aware Minimization (SAM) has shown promise in this area by reducing the sharpness of the loss landscape to obtain more generalized models. However, SAM and its variants sometimes fail to guide the model toward a flat minimum, and their training processes exhibit limitations, hindering further improvements in model generalization. In this paper, we first propose an improved model training process aimed at encouraging the model to converge to a flat minima. To achieve this, we design a curvature metric that has a minimal effect when the model is far from convergence but becomes increasingly influential in indicating the curvature of the minima as the model approaches a local minimum. Then we derive a novel algorithm from this metric, called Meta Curvature-Aware Minimization (MeCAM), to minimize the curvature around the local minima. Specifically, the optimization objective of MeCAM simultaneously minimizes the regular training loss, the surrogate gap of SAM, and the surrogate gap of meta-learning. We provide theoretical analysis on MeCAM's generalization error and convergence rate, and demonstrate its superiority over existing DG methods through extensive experiments on five benchmark DG datasets, including PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet. Code will be available on GitHub.
comment: 22 pages, 5 figures, 16 tables
♻ ☆ Closed-Loop Open-Vocabulary Mobile Manipulation with GPT-4V ICRA
Autonomous robot navigation and manipulation in open environments require reasoning and replanning with closed-loop feedback. In this work, we present COME-robot, the first closed-loop robotic system utilizing the GPT-4V vision-language foundation model for open-ended reasoning and adaptive planning in real-world scenarios.COME-robot incorporates two key innovative modules: (i) a multi-level open-vocabulary perception and situated reasoning module that enables effective exploration of the 3D environment and target object identification using commonsense knowledge and situated information, and (ii) an iterative closed-loop feedback and restoration mechanism that verifies task feasibility, monitors execution success, and traces failure causes across different modules for robust failure recovery. Through comprehensive experiments involving 8 challenging real-world mobile and tabletop manipulation tasks, COME-robot demonstrates a significant improvement in task success rate (~35%) compared to state-of-the-art methods. We further conduct comprehensive analyses to elucidate how COME-robot's design facilitates failure recovery, free-form instruction following, and long-horizon task planning.
comment: 6 pages, Accepted at 2025 IEEE ICRA, website: https://come-robot.github.io/
♻ ☆ FoundationStereo: Zero-Shot Stereo Matching CVPR 2025
Tremendous progress has been made in deep stereo matching to excel on benchmark datasets through per-domain fine-tuning. However, achieving strong zero-shot generalization - a hallmark of foundation models in other computer vision tasks - remains challenging for stereo matching. We introduce FoundationStereo, a foundation model for stereo depth estimation designed to achieve strong zero-shot generalization. To this end, we first construct a large-scale (1M stereo pairs) synthetic training dataset featuring large diversity and high photorealism, followed by an automatic self-curation pipeline to remove ambiguous samples. We then design a number of network architecture components to enhance scalability, including a side-tuning feature backbone that adapts rich monocular priors from vision foundation models to mitigate the sim-to-real gap, and long-range context reasoning for effective cost volume filtering. Together, these components lead to strong robustness and accuracy across domains, establishing a new standard in zero-shot stereo depth estimation. Project page: https://nvlabs.github.io/FoundationStereo/
comment: CVPR 2025
♻ ☆ Rethinking Pre-Trained Feature Extractor Selection in Multiple Instance Learning for Whole Slide Image Classification
Multiple instance learning (MIL) has become a preferred method for gigapixel whole slide image (WSI) classification without requiring patch-level annotations. Current MIL research primarily relies on embedding-based approaches, which extract patch features using a pre-trained feature extractor and aggregate them for slide-level prediction. Despite the critical role of feature extraction, there is limited guidance on selecting optimal feature extractors to maximize WSI performance. This study addresses this gap by systematically evaluating MIL feature extractors across three dimensions: pre-training dataset, backbone model, and pre-training method. Extensive experiments were conducted on two public WSI datasets (TCGA-NSCLC and Camelyon16) using four state-of-the-art (SOTA) MIL models. Our findings reveal that: 1) selecting a robust self-supervised learning (SSL) method has a greater impact on performance than relying solely on an in-domain pre-training dataset; 2) prioritizing Transformer-based backbones with deeper architectures over CNN-based models; and 3) using larger, more diverse pre-training datasets significantly enhances classification outcomes. We hope that these insights can provide practical guidance for optimizing WSI classification and explain the reasons behind the performance advantages of the current SOTA pathology foundation models. Furthermore, this work may inform the development of more effective pathology foundation models. Our code is publicly available at https://github.com/bryanwong17/MIL-Feature-Extractor-Selection
comment: Accepted to IEEE International Symposium on Biomedical Imaging (ISBI) 2025
♻ ☆ FastTrackTr:Towards Fast Multi-Object Tracking with Transformers
Transformer-based multi-object tracking (MOT) methods have captured the attention of many researchers in recent years. However, these models often suffer from slow inference speeds due to their structure or other issues. To address this problem, we revisited the Joint Detection and Tracking (JDT) method by looking back at past approaches. By integrating the original JDT approach with some advanced theories, this paper employs an efficient method of information transfer between frames on the DETR, constructing a fast and novel JDT-type MOT framework: FastTrackTr. Thanks to the superiority of this information transfer method, our approach not only reduces the number of queries required during tracking but also avoids the excessive introduction of network structures, ensuring model simplicity. Experimental results indicate that our method has the potential to achieve real-time tracking and exhibits competitive tracking accuracy across multiple datasets.
♻ ☆ FloNa: Floor Plan Guided Embodied Visual Navigation AAAI 2025
Humans naturally rely on floor plans to navigate in unfamiliar environments, as they are readily available, reliable, and provide rich geometrical guidance. However, existing visual navigation settings overlook this valuable prior knowledge, leading to limited efficiency and accuracy. To eliminate this gap, we introduce a novel navigation task: Floor Plan Visual Navigation (FloNa), the first attempt to incorporate floor plan into embodied visual navigation. While the floor plan offers significant advantages, two key challenges emerge: (1) handling the spatial inconsistency between the floor plan and the actual scene layout for collision-free navigation, and (2) aligning observed images with the floor plan sketch despite their distinct modalities. To address these challenges, we propose FloDiff, a novel diffusion policy framework incorporating a localization module to facilitate alignment between the current observation and the floor plan. We further collect $20k$ navigation episodes across $117$ scenes in the iGibson simulator to support the training and evaluation. Extensive experiments demonstrate the effectiveness and efficiency of our framework in unfamiliar scenes using floor plan knowledge. Project website: https://gauleejx.github.io/flona/.
comment: Accepted by AAAI 2025
♻ ☆ BuildingView: Constructing Urban Building Exteriors Databases with Street View Imagery and Multimodal Large Language Mode
Urban Building Exteriors are increasingly important in urban analytics, driven by advancements in Street View Imagery and its integration with urban research. Multimodal Large Language Models (LLMs) offer powerful tools for urban annotation, enabling deeper insights into urban environments. However, challenges remain in creating accurate and detailed urban building exterior databases, identifying critical indicators for energy efficiency, environmental sustainability, and human-centric design, and systematically organizing these indicators. To address these challenges, we propose BuildingView, a novel approach that integrates high-resolution visual data from Google Street View with spatial information from OpenStreetMap via the Overpass API. This research improves the accuracy of urban building exterior data, identifies key sustainability and design indicators, and develops a framework for their extraction and categorization. Our methodology includes a systematic literature review, building and Street View sampling, and annotation using the ChatGPT-4O API. The resulting database, validated with data from New York City, Amsterdam, and Singapore, provides a comprehensive tool for urban studies, supporting informed decision-making in urban planning, architectural design, and environmental policy. The code for BuildingView is available at https://github.com/Jasper0122/BuildingView.
comment: 15 pages, 6 figures
♻ ☆ InstaFace: Identity-Preserving Facial Editing with Single Image Inference
Facial appearance editing is crucial for digital avatars, AR/VR, and personalized content creation, driving realistic user experiences. However, preserving identity with generative models is challenging, especially in scenarios with limited data availability. Traditional methods often require multiple images and still struggle with unnatural face shifts, inconsistent hair alignment, or excessive smoothing effects. To overcome these challenges, we introduce a novel diffusion-based framework, InstaFace, to generate realistic images while preserving identity using only a single image. Central to InstaFace, we introduce an efficient guidance network that harnesses 3D perspectives by integrating multiple 3DMM-based conditionals without introducing additional trainable parameters. Moreover, to ensure maximum identity retention as well as preservation of background, hair, and other contextual features like accessories, we introduce a novel module that utilizes feature embeddings from a facial recognition model and a pre-trained vision-language model. Quantitative evaluations demonstrate that our method outperforms several state-of-the-art approaches in terms of identity preservation, photorealism, and effective control of pose, expression, and lighting.
♻ ☆ Attention Mechanism based Cognition-level Scene Understanding
Given a question-image input, the Visual Commonsense Reasoning (VCR) model can predict an answer with the corresponding rationale, which requires inference ability from the real world. The VCR task, which calls for exploiting the multi-source information as well as learning different levels of understanding and extensive commonsense knowledge, is a cognition-level scene understanding task. The VCR task has aroused researchers' interest due to its wide range of applications, including visual question answering, automated vehicle systems, and clinical decision support. Previous approaches to solving the VCR task generally rely on pre-training or exploiting memory with long dependency relationship encoded models. However, these approaches suffer from a lack of generalizability and losing information in long sequences. In this paper, we propose a parallel attention-based cognitive VCR network PAVCR, which fuses visual-textual information efficiently and encodes semantic information in parallel to enable the model to capture rich information for cognition-level inference. Extensive experiments show that the proposed model yields significant improvements over existing methods on the benchmark VCR dataset. Moreover, the proposed model provides intuitive interpretation into visual commonsense reasoning.
comment: Published in Information
♻ ☆ M2Distill: Multi-Modal Distillation for Lifelong Imitation Learning ICRA 2025
Lifelong imitation learning for manipulation tasks poses significant challenges due to distribution shifts that occur in incremental learning steps. Existing methods often focus on unsupervised skill discovery to construct an ever-growing skill library or distillation from multiple policies, which can lead to scalability issues as diverse manipulation tasks are continually introduced and may fail to ensure a consistent latent space throughout the learning process, leading to catastrophic forgetting of previously learned skills. In this paper, we introduce M2Distill, a multi-modal distillation-based method for lifelong imitation learning focusing on preserving consistent latent space across vision, language, and action distributions throughout the learning process. By regulating the shifts in latent representations across different modalities from previous to current steps, and reducing discrepancies in Gaussian Mixture Model (GMM) policies between consecutive learning steps, we ensure that the learned policy retains its ability to perform previously learned tasks while seamlessly integrating new skills. Extensive evaluations on the LIBERO lifelong imitation learning benchmark suites, including LIBERO-OBJECT, LIBERO-GOAL, and LIBERO-SPATIAL, demonstrate that our method consistently outperforms prior state-of-the-art methods across all evaluated metrics.
comment: IEEE ICRA 2025
♻ ☆ Jointly Understand Your Command and Intention:Reciprocal Co-Evolution between Scene-Aware 3D Human Motion Synthesis and Analysis
As two intimate reciprocal tasks, scene-aware human motion synthesis and analysis require a joint understanding between multiple modalities, including 3D body motions, 3D scenes, and textual descriptions. In this paper, we integrate these two paired processes into a Co-Evolving Synthesis-Analysis (CESA) pipeline and mutually benefit their learning. Specifically, scene-aware text-to-human synthesis generates diverse indoor motion samples from the same textual description to enrich human-scene interaction intra-class diversity, thus significantly benefiting training a robust human motion analysis system. Reciprocally, human motion analysis would enforce semantic scrutiny on each synthesized motion sample to ensure its semantic consistency with the given textual description, thus improving realistic motion synthesis. Considering that real-world indoor human motions are goal-oriented and path-guided, we propose a cascaded generation strategy that factorizes text-driven scene-specific human motion generation into three stages: goal inferring, path planning, and pose synthesizing. Coupling CESA with this powerful cascaded motion synthesis model, we jointly improve realistic human motion synthesis and robust human motion analysis in 3D scenes.
♻ ☆ Surgical-LVLM: Learning to Adapt Large Vision-Language Model for Grounded Visual Question Answering in Robotic Surgery ICLR 2025
Recent advancements in Surgical Visual Question Answering (Surgical-VQA) and related region grounding have shown great promise for robotic and medical applications, addressing the critical need for automated methods in personalized surgical mentorship. However, existing models primarily provide simple structured answers and struggle with complex scenarios due to their limited capability in recognizing long-range dependencies and aligning multimodal information. In this paper, we introduce Surgical-LVLM, a novel personalized large vision-language model tailored for complex surgical scenarios. Leveraging the pre-trained large vision-language model and specialized Visual Perception LoRA (VP-LoRA) blocks, our model excels in understanding complex visual-language tasks within surgical contexts. In addressing the visual grounding task, we propose the Token-Interaction (TIT) module, which strengthens the interaction between the grounding module and the language responses of the Large Visual Language Model (LVLM) after projecting them into the latent space. We demonstrate the effectiveness of Surgical-LVLM on several benchmarks, including EndoVis-17-VQLA, EndoVis-18-VQLA, and a newly introduced EndoVis Conversations dataset, which sets new performance standards. Our work contributes to advancing the field of automated surgical mentorship by providing a context-aware solution.
comment: The manuscript is accepted by ICLR 2025 FM-Wild Workshop
♻ ☆ Real-Time Convolutional Neural Network-Based Star Detection and Centroiding Method for CubeSat Star Tracker
Star trackers are one of the most accurate celestial sensors used for absolute attitude determination. The devices detect stars in captured images and accurately compute their projected centroids on an imaging focal plane with subpixel precision. Traditional algorithms for star detection and centroiding often rely on threshold adjustments for star pixel detection and pixel brightness weighting for centroid computation. However, challenges like high sensor noise and stray light can compromise algorithm performance. This article introduces a Convolutional Neural Network (CNN)-based approach for star detection and centroiding, tailored to address the issues posed by noisy star tracker images in the presence of stray light and other artifacts. Trained using simulated star images overlayed with real sensor noise and stray light, the CNN produces both a binary segmentation map distinguishing star pixels from the background and a distance map indicating each pixel's proximity to the nearest star centroid. Leveraging this distance information alongside pixel coordinates transforms centroid calculations into a set of trilateration problems solvable via the least squares method. Our method employs efficient UNet variants for the underlying CNN architectures, and the variants' performances are evaluated. Comprehensive testing has been undertaken with synthetic image evaluations, hardware-in-the-loop assessments, and night sky tests. The tests consistently demonstrated that our method outperforms several existing algorithms in centroiding accuracy and exhibits superior resilience to high sensor noise and stray light interference. An additional benefit of our algorithms is that they can be executed in real-time on low-power edge AI processors.
♻ ☆ RowDetr: End-to-End Row Detection Using Polynomials
Crop row detection is essential for enabling autonomous navigation in GPS-denied environments, such as under-canopy agricultural settings. Traditional methods often struggle with occlusions, variable lighting conditions, and the structural variability of crop rows. To address these challenges, RowDetr, a novel end-to-end neural network architecture, is introduced for robust and efficient row detection. A new dataset of approximately 6,900 images is curated, capturing a diverse range of real-world agricultural conditions, including occluded rows, uneven terrain, and varying crop densities. Unlike previous approaches, RowDetr leverages smooth polynomial functions to precisely delineate crop boundaries in the image space, ensuring a more structured and interpretable representation of row geometry. A key innovation of this approach is PolyOptLoss, a novel energy-based loss function designed to enhance learning robustness, even in the presence of noisy or imperfect labels. This loss function significantly improves model stability and generalization by optimizing polynomial curve fitting directly in image space. Extensive experiments demonstrate that RowDetr significantly outperforms existing frameworks, including Agronav and RowColAttention, across key performance metrics. Additionally, RowDetr achieves a sixfold speedup over Agronav, making it highly suitable for real-time deployment on resource-constrained edge devices. To facilitate better comparisons across future studies, lane detection metrics from autonomous driving research are adapted, providing a more standardized and meaningful evaluation framework for crop row detection. This work establishes a new benchmark in under-canopy
comment: Code will be open sourced upon publication
Machine Learning 151
☆ Multi-Fidelity Policy Gradient Algorithms
Many reinforcement learning (RL) algorithms require large amounts of data, prohibiting their use in applications where frequent interactions with operational systems are infeasible, or high-fidelity simulations are expensive or unavailable. Meanwhile, low-fidelity simulators--such as reduced-order models, heuristic reward functions, or generative world models--can cheaply provide useful data for RL training, even if they are too coarse for direct sim-to-real transfer. We propose multi-fidelity policy gradients (MFPGs), an RL framework that mixes a small amount of data from the target environment with a large volume of low-fidelity simulation data to form unbiased, reduced-variance estimators (control variates) for on-policy policy gradients. We instantiate the framework by developing multi-fidelity variants of two policy gradient algorithms: REINFORCE and proximal policy optimization. Experimental results across a suite of simulated robotics benchmark problems demonstrate that when target-environment samples are limited, MFPG achieves up to 3.9x higher reward and improves training stability when compared to baselines that only use high-fidelity data. Moreover, even when the baselines are given more high-fidelity samples--up to 10x as many interactions with the target environment--MFPG continues to match or outperform them. Finally, we observe that MFPG is capable of training effective policies even when the low-fidelity environment is drastically different from the target environment. MFPG thus not only offers a novel paradigm for efficient sim-to-real transfer but also provides a principled approach to managing the trade-off between policy performance and data collection costs.
☆ Fairness-Aware Low-Rank Adaptation Under Demographic Privacy Constraints
Pre-trained foundation models can be adapted for specific tasks using Low-Rank Adaptation (LoRA). However, the fairness properties of these adapted classifiers remain underexplored. Existing fairness-aware fine-tuning methods rely on direct access to sensitive attributes or their predictors, but in practice, these sensitive attributes are often held under strict consumer privacy controls, and neither the attributes nor their predictors are available to model developers, hampering the development of fair models. To address this issue, we introduce a set of LoRA-based fine-tuning methods that can be trained in a distributed fashion, where model developers and fairness auditors collaborate without sharing sensitive attributes or predictors. In this paper, we evaluate three such methods - sensitive unlearning, adversarial training, and orthogonality loss - against a fairness-unaware baseline, using experiments on the CelebA and UTK-Face datasets with an ImageNet pre-trained ViT-Base model. We find that orthogonality loss consistently reduces bias while maintaining or improving utility, whereas adversarial training improves False Positive Rate Parity and Demographic Parity in some cases, and sensitive unlearning provides no clear benefit. In tasks where significant biases are present, distributed fairness-aware fine-tuning methods can effectively eliminate bias without compromising consumer privacy and, in most cases, improve model utility.
☆ Understanding the Limits of Lifelong Knowledge Editing in LLMs
Keeping large language models factually up-to-date is crucial for deployment, yet costly retraining remains a challenge. Knowledge editing offers a promising alternative, but methods are only tested on small-scale or synthetic edit benchmarks. In this work, we aim to bridge research into lifelong knowledge editing to real-world edits at practically relevant scale. We first introduce WikiBigEdit; a large-scale benchmark of real-world Wikidata edits, built to automatically extend lifelong for future-proof benchmarking. In its first instance, it includes over 500K question-answer pairs for knowledge editing alongside a comprehensive evaluation pipeline. Finally, we use WikiBigEdit to study existing knowledge editing techniques' ability to incorporate large volumes of real-world facts and contrast their capabilities to generic modification techniques such as retrieval augmentation and continual finetuning to acquire a complete picture of the practical extent of current lifelong knowledge editing.
comment: Preprint
☆ Algorithmic Data Minimization for Machine Learning over Internet-of-Things Data Streams
Machine learning can analyze vast amounts of data generated by IoT devices to identify patterns, make predictions, and enable real-time decision-making. By processing sensor data, machine learning models can optimize processes, improve efficiency, and enhance personalized user experiences in smart systems. However, IoT systems are often deployed in sensitive environments such as households and offices, where they may inadvertently expose identifiable information, including location, habits, and personal identifiers. This raises significant privacy concerns, necessitating the application of data minimization -- a foundational principle in emerging data regulations, which mandates that service providers only collect data that is directly relevant and necessary for a specified purpose. Despite its importance, data minimization lacks a precise technical definition in the context of sensor data, where collections of weak signals make it challenging to apply a binary "relevant and necessary" rule. This paper provides a technical interpretation of data minimization in the context of sensor streams, explores practical methods for implementation, and addresses the challenges involved. Through our approach, we demonstrate that our framework can reduce user identifiability by up to 16.7% while maintaining accuracy loss below 1%, offering a viable path toward privacy-preserving IoT data processing.
comment: 9 pages, 18 figures
☆ AIM-Fair: Advancing Algorithmic Fairness via Selectively Fine-Tuning Biased Models with Contextual Synthetic Data CVPR 2025
Recent advances in generative models have sparked research on improving model fairness with AI-generated data. However, existing methods often face limitations in the diversity and quality of synthetic data, leading to compromised fairness and overall model accuracy. Moreover, many approaches rely on the availability of demographic group labels, which are often costly to annotate. This paper proposes AIM-Fair, aiming to overcome these limitations and harness the potential of cutting-edge generative models in promoting algorithmic fairness. We investigate a fine-tuning paradigm starting from a biased model initially trained on real-world data without demographic annotations. This model is then fine-tuned using unbiased synthetic data generated by a state-of-the-art diffusion model to improve its fairness. Two key challenges are identified in this fine-tuning paradigm, 1) the low quality of synthetic data, which can still happen even with advanced generative models, and 2) the domain and bias gap between real and synthetic data. To address the limitation of synthetic data quality, we propose Contextual Synthetic Data Generation (CSDG) to generate data using a text-to-image diffusion model (T2I) with prompts generated by a context-aware LLM, ensuring both data diversity and control of bias in synthetic data. To resolve domain and bias shifts, we introduce a novel selective fine-tuning scheme in which only model parameters more sensitive to bias and less sensitive to domain shift are updated. Experiments on CelebA and UTKFace datasets show that our AIM-Fair improves model fairness while maintaining utility, outperforming both fully and partially fine-tuned approaches to model fairness.
comment: Accepted at CVPR 2025. Github: https://github.com/zengqunzhao/AIM-Fair. Project page: https://zengqunzhao.github.io/AIMFair
☆ On Mitigating Affinity Bias through Bandits with Evolving Biased Feedback
Unconscious bias has been shown to influence how we assess our peers, with consequences for hiring, promotions and admissions. In this work, we focus on affinity bias, the component of unconscious bias which leads us to prefer people who are similar to us, despite no deliberate intention of favoritism. In a world where the people hired today become part of the hiring committee of tomorrow, we are particularly interested in understanding (and mitigating) how affinity bias affects this feedback loop. This problem has two distinctive features: 1) we only observe the biased value of a candidate, but we want to optimize with respect to their real value 2) the bias towards a candidate with a specific set of traits depends on the fraction of people in the hiring committee with the same set of traits. We introduce a new bandits variant that exhibits those two features, which we call affinity bandits. Unsurprisingly, classical algorithms such as UCB often fail to identify the best arm in this setting. We prove a new instance-dependent regret lower bound, which is larger than that in the standard bandit setting by a multiplicative function of $K$. Since we treat rewards that are time-varying and dependent on the policy's past actions, deriving this lower bound requires developing proof techniques beyond the standard bandit techniques. Finally, we design an elimination-style algorithm which nearly matches this regret, despite never observing the real rewards.
☆ NoT: Federated Unlearning via Weight Negation
Federated unlearning (FU) aims to remove a participant's data contributions from a trained federated learning (FL) model, ensuring privacy and regulatory compliance. Traditional FU methods often depend on auxiliary storage on either the client or server side or require direct access to the data targeted for removal-a dependency that may not be feasible if the data is no longer available. To overcome these limitations, we propose NoT, a novel and efficient FU algorithm based on weight negation (multiplying by -1), which circumvents the need for additional storage and access to the target data. We argue that effective and efficient unlearning can be achieved by perturbing model parameters away from the set of optimal parameters, yet being well-positioned for quick re-optimization. This technique, though seemingly contradictory, is theoretically grounded: we prove that the weight negation perturbation effectively disrupts inter-layer co-adaptation, inducing unlearning while preserving an approximate optimality property, thereby enabling rapid recovery. Experimental results across three datasets and three model architectures demonstrate that NoT significantly outperforms existing baselines in unlearning efficacy as well as in communication and computational efficiency.
comment: The 42nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville TN, US. 2025
☆ BEHAVIOR Robot Suite: Streamlining Real-World Whole-Body Manipulation for Everyday Household Activities
Real-world household tasks present significant challenges for mobile manipulation robots. An analysis of existing robotics benchmarks reveals that successful task performance hinges on three key whole-body control capabilities: bimanual coordination, stable and precise navigation, and extensive end-effector reachability. Achieving these capabilities requires careful hardware design, but the resulting system complexity further complicates visuomotor policy learning. To address these challenges, we introduce the BEHAVIOR Robot Suite (BRS), a comprehensive framework for whole-body manipulation in diverse household tasks. Built on a bimanual, wheeled robot with a 4-DoF torso, BRS integrates a cost-effective whole-body teleoperation interface for data collection and a novel algorithm for learning whole-body visuomotor policies. We evaluate BRS on five challenging household tasks that not only emphasize the three core capabilities but also introduce additional complexities, such as long-range navigation, interaction with articulated and deformable objects, and manipulation in confined spaces. We believe that BRS's integrated robotic embodiment, data collection interface, and learning framework mark a significant step toward enabling real-world whole-body manipulation for everyday household tasks. BRS is open-sourced at https://behavior-robot-suite.github.io/
comment: Project website: https://behavior-robot-suite.github.io/
☆ Physics-based machine learning framework for predicting NOx emissions from compression ignition engines using on-board diagnostics data
This work presents a physics-based machine learning framework to predict and analyze oxides of nitrogen (NOx) emissions from compression-ignition engine-powered vehicles using on-board diagnostics (OBD) data as input. Accurate NOx prediction from OBD datasets is difficult because NOx formation inside an engine combustion chamber is governed by complex processes occurring on timescales much shorter than the data collection rate. Thus, emissions generally cannot be predicted accurately using simple empirically derived physics models. Black box models like genetic algorithms or neural networks can be more accurate, but have poor interpretability. The transparent model presented in this paper has both high accuracy and can explain potential sources of high emissions. The proposed framework consists of two major steps: a physics-based NOx prediction model combined with a novel Divergent Window Co-occurrence (DWC) Pattern detection algorithm to analyze operating conditions that are not adequately addressed by the physics-based model. The proposed framework is validated for generalizability with a second vehicle OBD dataset, a sensitivity analysis is performed, and model predictions are compared with that from a deep neural network. The results show that NOx emissions predictions using the proposed model has around 55% better root mean square error, and around 60% higher mean absolute error compared to the baseline NOx prediction model from previously published work. The DWC Pattern Detection Algorithm identified low engine power conditions to have high statistical significance, indicating an operating regime where the model can be improved. This work shows that the physics-based machine learning framework is a viable method for predicting NOx emissions from engines that do not incorporate NOx sensing.
☆ Symbolic Mixture-of-Experts: Adaptive Skill-based Routing for Heterogeneous Reasoning
Combining existing pre-trained expert LLMs is a promising avenue for scalably tackling large-scale and diverse tasks. However, selecting experts at the task level is often too coarse-grained, as heterogeneous tasks may require different expertise for each instance. To enable adaptive instance-level mixing of pre-trained LLM experts, we propose Symbolic-MoE, a symbolic, text-based, and gradient-free Mixture-of-Experts framework. Symbolic-MoE takes a fine-grained approach to selection by emphasizing skills, e.g., algebra in math or molecular biology in biomedical reasoning. We propose a skill-based recruiting strategy that dynamically selects the most relevant set of expert LLMs for diverse reasoning tasks based on their strengths. Each selected expert then generates its own reasoning, resulting in k outputs from k experts, which are then synthesized into a final high-quality response by an aggregator chosen based on its ability to integrate diverse reasoning outputs. We show that Symbolic-MoE's instance-level expert selection improves performance by a large margin but -- when implemented naively -- can introduce a high computational overhead due to the need for constant model loading and offloading. To address this, we implement a batch inference strategy that groups instances based on their assigned experts, loading each model only once. This allows us to integrate 16 expert models on 1 GPU with a time cost comparable to or better than prior multi-agent baselines using 4 GPUs. Through extensive evaluations on diverse benchmarks (MMLU-Pro, GPQA, AIME, and MedMCQA), we demonstrate that Symbolic-MoE outperforms strong LLMs like GPT4o-mini, as well as multi-agent approaches, with an absolute average improvement of 8.15% over the best multi-agent baseline. Moreover, Symbolic-MoE removes the need for expensive multi-round discussions, outperforming discussion baselines with less computation.
comment: The first three authors contributed equally. Project Page: https://symbolic_moe.github.io/
☆ Strategy Coopetition Explains the Emergence and Transience of In-Context Learning
In-context learning (ICL) is a powerful ability that emerges in transformer models, enabling them to learn from context without weight updates. Recent work has established emergent ICL as a transient phenomenon that can sometimes disappear after long training times. In this work, we sought a mechanistic understanding of these transient dynamics. Firstly, we find that, after the disappearance of ICL, the asymptotic strategy is a remarkable hybrid between in-weights and in-context learning, which we term "context-constrained in-weights learning" (CIWL). CIWL is in competition with ICL, and eventually replaces it as the dominant strategy of the model (thus leading to ICL transience). However, we also find that the two competing strategies actually share sub-circuits, which gives rise to cooperative dynamics as well. For example, in our setup, ICL is unable to emerge quickly on its own, and can only be enabled through the simultaneous slow development of asymptotic CIWL. CIWL thus both cooperates and competes with ICL, a phenomenon we term "strategy coopetition." We propose a minimal mathematical model that reproduces these key dynamics and interactions. Informed by this model, we were able to identify a setup where ICL is truly emergent and persistent.
comment: 20 pages, 18 figures
☆ Decision-aware training of spatiotemporal forecasting models
Optimal allocation of scarce resources is a common problem for decision makers faced with choosing a limited number of locations for intervention. Spatiotemporal prediction models could make such decisions data-driven. A recent performance metric called fraction of best possible reach (BPR) measures the impact of using a model's recommended size K subset of sites compared to the best possible top-K in hindsight. We tackle two open problems related to BPR. First, we explore how to rank all sites numerically given a probabilistic model that predicts event counts jointly across sites. Ranking via the per-site mean is suboptimal for BPR. Instead, we offer a better ranking for BPR backed by decision theory. Second, we explore how to train a probabilistic model's parameters to maximize BPR. Discrete selection of K sites implies all-zero parameter gradients which prevent standard gradient training. We overcome this barrier via advances in perturbed optimizers. We further suggest a training objective that combines likelihood with a decision-aware BPR constraint to deliver high-quality top-K rankings as well as good forecasts for all sites. We demonstrate our approach on two where-to-intervene applications: mitigating opioid-related fatal overdoses for public health and monitoring endangered wildlife.
comment: 9 pages, 3 figures
☆ Conformal Prediction for Image Segmentation Using Morphological Prediction Sets
Image segmentation is a challenging task influenced by multiple sources of uncertainty, such as the data labeling process or the sampling of training data. In this paper we focus on binary segmentation and address these challenges using conformal prediction, a family of model- and data-agnostic methods for uncertainty quantification that provide finite-sample theoretical guarantees and applicable to any pretrained predictor. Our approach involves computing nonconformity scores, a type of prediction residual, on held-out calibration data not used during training. We use dilation, one of the fundamental operations in mathematical morphology, to construct a margin added to the borders of predicted segmentation masks. At inference, the predicted set formed by the mask and its margin contains the ground-truth mask with high probability, at a confidence level specified by the user. The size of the margin serves as an indicator of predictive uncertainty for a given model and dataset. We work in a regime of minimal information as we do not require any feedback from the predictor: only the predicted masks are needed for computing the prediction sets. Hence, our method is applicable to any segmentation model, including those based on deep learning; we evaluate our approach on several medical imaging applications.
☆ Can KAN CANs? Input-convex Kolmogorov-Arnold Networks (KANs) as hyperelastic constitutive artificial neural networks (CANs)
Traditional constitutive models rely on hand-crafted parametric forms with limited expressivity and generalizability, while neural network-based models can capture complex material behavior but often lack interpretability. To balance these trade-offs, we present Input-Convex Kolmogorov-Arnold Networks (ICKANs) for learning polyconvex hyperelastic constitutive laws. ICKANs leverage the Kolmogorov-Arnold representation, decomposing the model into compositions of trainable univariate spline-based activation functions for rich expressivity. We introduce trainable input-convex splines within the KAN architecture, ensuring physically admissible polyconvex hyperelastic models. The resulting models are both compact and interpretable, enabling explicit extraction of analytical constitutive relationships through an input-convex symbolic regression techinque. Through unsupervised training on full-field strain data and limited global force measurements, ICKANs accurately capture nonlinear stress-strain behavior across diverse strain states. Finite element simulations of unseen geometries with trained ICKAN hyperelastic constitutive models confirm the framework's robustness and generalization capability.
comment: 34 pages, 15 figures
☆ A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models
Large Language Models (LLMs) have revolutionized natural language processing, yet their internal mechanisms remain largely opaque. Recently, mechanistic interpretability has attracted significant attention from the research community as a means to understand the inner workings of LLMs. Among various mechanistic interpretability approaches, Sparse Autoencoders (SAEs) have emerged as a particularly promising method due to their ability to disentangle the complex, superimposed features within LLMs into more interpretable components. This paper presents a comprehensive examination of SAEs as a promising approach to interpreting and understanding LLMs. We provide a systematic overview of SAE principles, architectures, and applications specifically tailored for LLM analysis, covering theoretical foundations, implementation strategies, and recent developments in sparsity mechanisms. We also explore how SAEs can be leveraged to explain the internal workings of LLMs, steer model behaviors in desired directions, and develop more transparent training methodologies for future models. Despite the challenges that remain around SAE implementation and scaling, they continue to provide valuable tools for understanding the internal mechanisms of large language models.
comment: 20 pages, 3 figures
☆ On the similarity of bandwidth-tuned quantum kernels and classical kernels
Quantum kernels (QK) are widely used in quantum machine learning applications; yet, their potential to surpass classical machine learning methods on classical datasets remains uncertain. This limitation can be attributed to the exponential concentration phenomenon, which can impair both trainability and generalization. A common strategy to alleviate this is bandwidth tuning, which involves rescaling data points in the quantum model to improve generalization. In this work, we numerically demonstrate that optimal bandwidth tuning results in QKs that closely resemble radial basis function (RBF) kernels, leading to a lack of quantum advantage over classical methods. Moreover, we reveal that the size of optimal bandwidth tuning parameters further simplifies QKs, causing them to behave like polynomial kernels, corresponding to a low-order Taylor approximation of a RBF kernel. We thoroughly investigate this for fidelity quantum kernels and projected quantum kernels using various data encoding circuits across several classification datasets. We provide numerical evidence and derive a simple analytical model that elucidates how bandwidth tuning influences key quantities in classification tasks. Overall, our findings shed light on the mechanisms that render QK methods classically simulatable.
comment: 9 main pages with 5 figures, and 9 appendix pages with 12 figures
☆ From Theory to Application: A Practical Introduction to Neural Operators in Scientific Computing
This focused review explores a range of neural operator architectures for approximating solutions to parametric partial differential equations (PDEs), emphasizing high-level concepts and practical implementation strategies. The study covers foundational models such as Deep Operator Networks (DeepONet), Principal Component Analysis-based Neural Networks (PCANet), and Fourier Neural Operators (FNO), providing comparative insights into their core methodologies and performance. These architectures are demonstrated on two classical linear parametric PDEs: the Poisson equation and linear elastic deformation. Beyond forward problem-solving, the review delves into applying neural operators as surrogates in Bayesian inference problems, showcasing their effectiveness in accelerating posterior inference while maintaining accuracy. The paper concludes by discussing current challenges, particularly in controlling prediction accuracy and generalization. It outlines emerging strategies to address these issues, such as residual-based error correction and multi-level training. This review can be seen as a comprehensive guide to implementing neural operators and integrating them into scientific computing workflows.
comment: 53 pages, 17 figures, Github repository: https://github.com/CEADpx/neural_operators
☆ Quantifying the Robustness of Retrieval-Augmented Language Models Against Spurious Features in Grounding Data
Robustness has become a critical attribute for the deployment of RAG systems in real-world applications. Existing research focuses on robustness to explicit noise (e.g., document semantics) but overlooks spurious features (a.k.a. implicit noise). While previous works have explored spurious features in LLMs, they are limited to specific features (e.g., formats) and narrow scenarios (e.g., ICL). In this work, we statistically confirm the presence of spurious features in the RAG paradigm, a robustness problem caused by the sensitivity of LLMs to semantic-agnostic features. Moreover, we provide a comprehensive taxonomy of spurious features and empirically quantify their impact through controlled experiments. Further analysis reveals that not all spurious features are harmful and they can even be beneficial sometimes. Extensive evaluation results across multiple LLMs suggest that spurious features are a widespread and challenging problem in the field of RAG. The code and dataset will be released to facilitate future research. We release all codes and data at: $\\\href{https://github.com/maybenotime/RAG-SpuriousFeatures}{https://github.com/maybenotime/RAG-SpuriousFeatures}$.
☆ MPTSNet: Integrating Multiscale Periodic Local Patterns and Global Dependencies for Multivariate Time Series Classification AAAI2025
Multivariate Time Series Classification (MTSC) is crucial in extensive practical applications, such as environmental monitoring, medical EEG analysis, and action recognition. Real-world time series datasets typically exhibit complex dynamics. To capture this complexity, RNN-based, CNN-based, Transformer-based, and hybrid models have been proposed. Unfortunately, current deep learning-based methods often neglect the simultaneous construction of local features and global dependencies at different time scales, lacking sufficient feature extraction capabilities to achieve satisfactory classification accuracy. To address these challenges, we propose a novel Multiscale Periodic Time Series Network (MPTSNet), which integrates multiscale local patterns and global correlations to fully exploit the inherent information in time series. Recognizing the multi-periodicity and complex variable correlations in time series, we use the Fourier transform to extract primary periods, enabling us to decompose data into multiscale periodic segments. Leveraging the inherent strengths of CNN and attention mechanism, we introduce the PeriodicBlock, which adaptively captures local patterns and global dependencies while offering enhanced interpretability through attention integration across different periodic scales. The experiments on UEA benchmark datasets demonstrate that the proposed MPTSNet outperforms 21 existing advanced baselines in the MTSC tasks.
comment: Accepted by AAAI2025
☆ opXRD: Open Experimental Powder X-ray Diffraction Database
Powder X-ray diffraction (pXRD) experiments are a cornerstone for materials structure characterization. Despite their widespread application, analyzing pXRD diffractograms still presents a significant challenge to automation and a bottleneck in high-throughput discovery in self-driving labs. Machine learning promises to resolve this bottleneck by enabling automated powder diffraction analysis. A notable difficulty in applying machine learning to this domain is the lack of sufficiently sized experimental datasets, which has constrained researchers to train primarily on simulated data. However, models trained on simulated pXRD patterns showed limited generalization to experimental patterns, particularly for low-quality experimental patterns with high noise levels and elevated backgrounds. With the Open Experimental Powder X-Ray Diffraction Database (opXRD), we provide an openly available and easily accessible dataset of labeled and unlabeled experimental powder diffractograms. Labeled opXRD data can be used to evaluate the performance of models on experimental data and unlabeled opXRD data can help improve the performance of models on experimental data, e.g. through transfer learning methods. We collected \numpatterns diffractograms, 2179 of them labeled, from a wide spectrum of materials classes. We hope this ongoing effort can guide machine learning research toward fully automated analysis of pXRD data and thus enable future self-driving materials labs.
☆ BARK: A Fully Bayesian Tree Kernel for Black-box Optimization
We perform Bayesian optimization using a Gaussian process perspective on Bayesian Additive Regression Trees (BART). Our BART Kernel (BARK) uses tree agreement to define a posterior over piecewise-constant functions, and we explore the space of tree kernels using a Markov chain Monte Carlo approach. Where BART only samples functions, the resulting BARK model obtains samples of Gaussian processes defining distributions over functions, which allow us to build acquisition functions for Bayesian optimization. Our tree-based approach enables global optimization over the surrogate, even for mixed-feature spaces. Moreover, where many previous tree-based kernels provide uncertainty quantification over function values, our sampling scheme captures uncertainty over the tree structure itself. Our experiments show the strong performance of BARK on both synthetic and applied benchmarks, due to the combination of our fully Bayesian surrogate and the optimization procedure.
comment: 8 main pages, 22 total pages, 10 figures, 6 tables
☆ InDRiVE: Intrinsic Disagreement based Reinforcement for Vehicle Exploration through Curiosity Driven Generalized World Model IROS 2025
Model-based Reinforcement Learning (MBRL) has emerged as a promising paradigm for autonomous driving, where data efficiency and robustness are critical. Yet, existing solutions often rely on carefully crafted, task specific extrinsic rewards, limiting generalization to new tasks or environments. In this paper, we propose InDRiVE (Intrinsic Disagreement based Reinforcement for Vehicle Exploration), a method that leverages purely intrinsic, disagreement based rewards within a Dreamer based MBRL framework. By training an ensemble of world models, the agent actively explores high uncertainty regions of environments without any task specific feedback. This approach yields a task agnostic latent representation, allowing for rapid zero shot or few shot fine tuning on downstream driving tasks such as lane following and collision avoidance. Experimental results in both seen and unseen environments demonstrate that InDRiVE achieves higher success rates and fewer infractions compared to DreamerV2 and DreamerV3 baselines despite using significantly fewer training steps. Our findings highlight the effectiveness of purely intrinsic exploration for learning robust vehicle control behaviors, paving the way for more scalable and adaptable autonomous driving systems.
comment: This work has been submitted to IROS 2025 and is currently under review
☆ Tractable Representations for Convergent Approximation of Distributional HJB Equations
In reinforcement learning (RL), the long-term behavior of decision-making policies is evaluated based on their average returns. Distributional RL has emerged, presenting techniques for learning return distributions, which provide additional statistics for evaluating policies, incorporating risk-sensitive considerations. When the passage of time cannot naturally be divided into discrete time increments, researchers have studied the continuous-time RL (CTRL) problem, where agent states and decisions evolve continuously. In this setting, the Hamilton-Jacobi-Bellman (HJB) equation is well established as the characterization of the expected return, and many solution methods exist. However, the study of distributional RL in the continuous-time setting is in its infancy. Recent work has established a distributional HJB (DHJB) equation, providing the first characterization of return distributions in CTRL. These equations and their solutions are intractable to solve and represent exactly, requiring novel approximation techniques. This work takes strides towards this end, establishing conditions on the method of parameterizing return distributions under which the DHJB equation can be approximately solved. Particularly, we show that under a certain topological property of the mapping between statistics learned by a distributional RL algorithm and corresponding distributions, approximation of these statistics leads to close approximations of the solution of the DHJB equation. Concretely, we demonstrate that the quantile representation common in distributional RL satisfies this topological property, certifying an efficient approximation algorithm for continuous-time distributional RL.
comment: Accepted to RLDM 2025
☆ Global graph features unveiled by unsupervised geometric deep learning
Graphs provide a powerful framework for modeling complex systems, but their structural variability makes analysis and classification challenging. To address this, we introduce GAUDI (Graph Autoencoder Uncovering Descriptive Information), a novel unsupervised geometric deep learning framework that captures both local details and global structure. GAUDI employs an innovative hourglass architecture with hierarchical pooling and upsampling layers, linked through skip connections to preserve essential connectivity information throughout the encoding-decoding process. By mapping different realizations of a system - generated from the same underlying parameters - into a continuous, structured latent space, GAUDI disentangles invariant process-level features from stochastic noise. We demonstrate its power across multiple applications, including modeling small-world networks, characterizing protein assemblies from super-resolution microscopy, analyzing collective motion in the Vicsek model, and capturing age-related changes in brain connectivity. This approach not only improves the analysis of complex graphs but also provides new insights into emergent phenomena across diverse scientific domains.
comment: 23 pages, 5 figures
☆ Diffusion Models for Cayley Graphs
We review the problem of finding paths in Cayley graphs of groups and group actions, using the Rubik's cube as an example, and we list several more examples of significant mathematical interest. We then show how to formulate these problems in the framework of diffusion models. The exploration of the graph is carried out by the forward process, while finding the target nodes is done by the inverse backward process. This systematizes the discussion and suggests many generalizations. To improve exploration, we propose a ``reversed score'' ansatz which substantially improves over previous comparable algorithms.
comment: 25 pages, 5 figures
☆ Revitalizing Saturated Benchmarks: A Weighted Metric Approach for Differentiating Large Language Model Performance NAACL
Existing benchmarks are becoming saturated and struggle to separate model performances due to factors like data contamination and advancing LLM capabilities. This paper introduces EMDM (Enhanced Model Differentiation Metric), a novel weighted metric that revitalizes benchmarks by enhancing model separation. EMDM integrates final answer and Chain-of-Thought (CoT) reasoning correctness, assigning weights based on the complexity and reasoning depth required to solve a given sample in the evaluation data. Using a baseline LLM in two setups-Unguided, where the model has no prior exposure to test samples, and Guided, where the model has prior knowledge of the desired answer-EMDM distinguishes instances of varying difficulty. The CoT and answer correctness from these setups inform an optimization objective for weight assignment, resulting in a more nuanced evaluation of model performance. Compared to the exact match (EM) metric, which achieves 17% separation on ARC-Challenge, EMDM achieves 46%, demonstrating its effectiveness in differentiating models based on reasoning and knowledge requirements.
comment: conference NAACL, TrustNLP Workshop
☆ Impoola: The Power of Average Pooling for Image-Based Deep Reinforcement Learning
As image-based deep reinforcement learning tackles more challenging tasks, increasing model size has become an important factor in improving performance. Recent studies achieved this by focusing on the parameter efficiency of scaled networks, typically using Impala-CNN, a 15-layer ResNet-inspired network, as the image encoder. However, while Impala-CNN evidently outperforms older CNN architectures, potential advancements in network design for deep reinforcement learning-specific image encoders remain largely unexplored. We find that replacing the flattening of output feature maps in Impala-CNN with global average pooling leads to a notable performance improvement. This approach outperforms larger and more complex models in the Procgen Benchmark, particularly in terms of generalization. We call our proposed encoder model Impoola-CNN. A decrease in the network's translation sensitivity may be central to this improvement, as we observe the most significant gains in games without agent-centered observations. Our results demonstrate that network scaling is not just about increasing model size - efficient network design is also an essential factor.
☆ Riemann$^2$: Learning Riemannian Submanifolds from Riemannian Data AISTATS 2025
Latent variable models are powerful tools for learning low-dimensional manifolds from high-dimensional data. However, when dealing with constrained data such as unit-norm vectors or symmetric positive-definite matrices, existing approaches ignore the underlying geometric constraints or fail to provide meaningful metrics in the latent space. To address these limitations, we propose to learn Riemannian latent representations of such geometric data. To do so, we estimate the pullback metric induced by a Wrapped Gaussian Process Latent Variable Model, which explicitly accounts for the data geometry. This enables us to define geometry-aware notions of distance and shortest paths in the latent space, while ensuring that our model only assigns probability mass to the data manifold. This generalizes previous work and allows us to handle complex tasks in various domains, including robot motion synthesis and analysis of brain connectomes.
comment: Accepted at AISTATS 2025
☆ Additive Model Boosting: New Insights and Path(ologie)s
Additive models (AMs) have sparked a lot of interest in machine learning recently, allowing the incorporation of interpretable structures into a wide range of model classes. Many commonly used approaches to fit a wide variety of potentially complex additive models build on the idea of boosting additive models. While boosted additive models (BAMs) work well in practice, certain theoretical aspects are still poorly understood, including general convergence behavior and what optimization problem is being solved when accounting for the implicit regularizing nature of boosting. In this work, we study the solution paths of BAMs and establish connections with other approaches for certain classes of problems. Along these lines, we derive novel convergence results for BAMs, which yield crucial insights into the inner workings of the method. While our results generally provide reassuring theoretical evidence for the practical use of BAMs, they also uncover some ``pathologies'' of boosting for certain additive model classes concerning their convergence behavior that require caution in practice. We empirically validate our theoretical findings through several numerical experiments.
☆ Leveraging Approximate Caching for Faster Retrieval-Augmented Generation
Retrieval-augmented generation (RAG) enhances the reliability of large language model (LLM) answers by integrating external knowledge. However, RAG increases the end-to-end inference time since looking for relevant documents from large vector databases is computationally expensive. To address this, we introduce Proximity, an approximate key-value cache that optimizes the RAG workflow by leveraging similarities in user queries. Instead of treating each query independently, Proximity reuses previously retrieved documents when similar queries appear, reducing reliance on expensive vector database lookups. We evaluate Proximity on the MMLU and MedRAG benchmarks, demonstrating that it significantly improves retrieval efficiency while maintaining response accuracy. Proximity reduces retrieval latency by up to 59% while maintaining accuracy and lowers the computational burden on the vector database. We also experiment with different similarity thresholds and quantify the trade-off between speed and recall. Our work shows that approximate caching is a viable and effective strategy for optimizing RAG-based systems.
☆ Machine Learning for Improved Density Functional Theory Thermodynamics
The predictive accuracy of density functional theory (DFT) for alloy formation enthalpies is often limited by intrinsic energy resolution errors, particularly in ternary phase stability calculations. In this work, we present a machine learning (ML) approach to systematically correct these errors, improving the reliability of first-principles predictions. A neural network model has been trained to predict the discrepancy between DFT-calculated and experimentally measured enthalpies for binary and ternary alloys and compounds. The model utilizes a structured feature set comprising elemental concentrations, atomic numbers, and interaction terms to capture key chemical and structural effects. By applying supervised learning and rigorous data curation we ensure a robust and physically meaningful correction. The model is implemented as a multi-layer perceptron (MLP) regressor with three hidden layers, optimized through leave-one-out cross-validation (LOOCV) and k-fold cross-validation to prevent overfitting. We illustrate the effectiveness of this method by applying it to the Al-Ni-Pd and Al-Ni-Ti systems, which are of interest for high-temperature applications in aerospace and protective coatings.
comment: 9 pages, 5 figures, 1 table
☆ Post-Hoc Concept Disentanglement: From Correlated to Isolated Concept Representations
Concept Activation Vectors (CAVs) are widely used to model human-understandable concepts as directions within the latent space of neural networks. They are trained by identifying directions from the activations of concept samples to those of non-concept samples. However, this method often produces similar, non-orthogonal directions for correlated concepts, such as "beard" and "necktie" within the CelebA dataset, which frequently co-occur in images of men. This entanglement complicates the interpretation of concepts in isolation and can lead to undesired effects in CAV applications, such as activation steering. To address this issue, we introduce a post-hoc concept disentanglement method that employs a non-orthogonality loss, facilitating the identification of orthogonal concept directions while preserving directional correctness. We evaluate our approach with real-world and controlled correlated concepts in CelebA and a synthetic FunnyBirds dataset with VGG16 and ResNet18 architectures. We further demonstrate the superiority of orthogonalized concept representations in activation steering tasks, allowing (1) the insertion of isolated concepts into input images through generative models and (2) the removal of concepts for effective shortcut suppression with reduced impact on correlated concepts in comparison to baseline CAVs.
☆ Removing Geometric Bias in One-Class Anomaly Detection with Adaptive Feature Perturbation WACV 2025
One-class anomaly detection aims to detect objects that do not belong to a predefined normal class. In practice training data lack those anomalous samples; hence state-of-the-art methods are trained to discriminate between normal and synthetically-generated pseudo-anomalous data. Most methods use data augmentation techniques on normal images to simulate anomalies. However the best-performing ones implicitly leverage a geometric bias present in the benchmarking datasets. This limits their usability in more general conditions. Others are relying on basic noising schemes that may be suboptimal in capturing the underlying structure of normal data. In addition most still favour the image domain to generate pseudo-anomalies training models end-to-end from only the normal class and overlooking richer representations of the information. To overcome these limitations we consider frozen yet rich feature spaces given by pretrained models and create pseudo-anomalous features with a novel adaptive linear feature perturbation technique. It adapts the noise distribution to each sample applies decaying linear perturbations to feature vectors and further guides the classification process using a contrastive learning objective. Experimental evaluation conducted on both standard and geometric bias-free datasets demonstrates the superiority of our approach with respect to comparable baselines. The codebase is accessible via our public repository.
comment: Published in WACV 2025
☆ Mol-CADiff: Causality-Aware Autoregressive Diffusion for Molecule Generation
The design of novel molecules with desired properties is a key challenge in drug discovery and materials science. Traditional methods rely on trial-and-error, while recent deep learning approaches have accelerated molecular generation. However, existing models struggle with generating molecules based on specific textual descriptions. We introduce Mol-CADiff, a novel diffusion-based framework that uses causal attention mechanisms for text-conditional molecular generation. Our approach explicitly models the causal relationship between textual prompts and molecular structures, overcoming key limitations in existing methods. We enhance dependency modeling both within and across modalities, enabling precise control over the generation process. Our extensive experiments demonstrate that Mol-CADiff outperforms state-of-the-art methods in generating diverse, novel, and chemically valid molecules, with better alignment to specified properties, enabling more intuitive language-driven molecular design.
☆ Statistical Deficiency for Task Inclusion Estimation
Tasks are central in machine learning, as they are the most natural objects to assess the capabilities of current models. The trend is to build general models able to address any task. Even though transfer learning and multitask learning try to leverage the underlying task space, no well-founded tools are available to study its structure. This study proposes a theoretically grounded setup to define the notion of task and to compute the {\bf inclusion} between two tasks from a statistical deficiency point of view. We propose a tractable proxy as information sufficiency to estimate the degree of inclusion between tasks, show its soundness on synthetic data, and use it to reconstruct empirically the classic NLP pipeline.
comment: 34 pages
☆ Bridging the Semantic Gap in Virtual Machine Introspection and Forensic Memory Analysis
Forensic Memory Analysis (FMA) and Virtual Machine Introspection (VMI) are critical tools for security in a virtualization-based approach. VMI and FMA involves using digital forensic methods to extract information from the system to identify and explain security incidents. A key challenge in both FMA and VMI is the "Semantic Gap", which is the difficulty of interpreting raw memory data without specialized tools and expertise. In this work, we investigate how a priori knowledge, metadata and engineered features can aid VMI and FMA, leveraging machine learning to automate information extraction and reduce the workload of forensic investigators. We choose OpenSSH as our use case to test different methods to extract high level structures. We also test our method on complete physical memory dumps to showcase the effectiveness of the engineered features. Our features range from basic statistical features to advanced graph-based representations using malloc headers and pointer translations. The training and testing are carried out on public datasets that we compare against already recognized baseline methods. We show that using metadata, we can improve the performance of the algorithm when there is very little training data and also quantify how having more data results in better generalization performance. The final contribution is an open dataset of physical memory dumps, totalling more than 1 TB of different memory state, software environments, main memory capacities and operating system versions. Our methods show that having more metadata boosts performance with all methods obtaining an F1-Score of over 80%. Our research underscores the possibility of using feature engineering and machine learning techniques to bridge the semantic gap.
☆ Enhancing Network Security: A Hybrid Approach for Detection and Mitigation of Distributed Denial-of-Service Attacks Using Machine Learning
The distributed denial-of-service (DDoS) attack stands out as a highly formidable cyber threat, representing an advanced form of the denial-of-service (DoS) attack. A DDoS attack involves multiple computers working together to overwhelm a system, making it unavailable. On the other hand, a DoS attack is a one-on-one attempt to make a system or website inaccessible. Thus, it is crucial to construct an effective model for identifying various DDoS incidents. Although extensive research has focused on binary detection models for DDoS identification, they face challenges to adapt evolving threats, necessitating frequent updates. Whereas multiclass detection models offer a comprehensive defense against diverse DDoS attacks, ensuring adaptability in the ever-changing cyber threat landscape. In this paper, we propose a Hybrid Model to strengthen network security by combining the featureextraction abilities of 1D Convolutional Neural Networks (CNNs) with the classification skills of Random Forest (RF) and Multi-layer Perceptron (MLP) classifiers. Using the CIC-DDoS2019 dataset, we perform multiclass classification of various DDoS attacks and conduct a comparative analysis of evaluation metrics for RF, MLP, and our proposed Hybrid Model. After analyzing the results, we draw meaningful conclusions and confirm the superiority of our Hybrid Model by performing thorough cross-validation. Additionally, we integrate our machine learning model with Snort, which provides a robust and adaptive solution for detecting and mitigating various DDoS attacks.
comment: Part of the book series: Communications in Computer and Information Science ((CCIS,volume 2091))
☆ Personalized Federated Learning via Learning Dynamic Graphs
Personalized Federated Learning (PFL) aims to train a personalized model for each client that is tailored to its local data distribution, learning fails to perform well on individual clients due to variations in their local data distributions. Most existing PFL methods focus on personalizing the aggregated global model for each client, neglecting the fundamental aspect of federated learning: the regulation of how client models are aggregated. Additionally, almost all of them overlook the graph structure formed by clients in federated learning. In this paper, we propose a novel method, Personalized Federated Learning with Graph Attention Network (pFedGAT), which captures the latent graph structure between clients and dynamically determines the importance of other clients for each client, enabling fine-grained control over the aggregation process. We evaluate pFedGAT across multiple data distribution scenarios, comparing it with twelve state of the art methods on three datasets: Fashion MNIST, CIFAR-10, and CIFAR-100, and find that it consistently performs well.
☆ Soft Policy Optimization: Online Off-Policy RL for Sequence Models
RL-based post-training of language models is almost exclusively done using on-policy methods such as PPO. These methods cannot learn from arbitrary sequences such as those produced earlier in training, in earlier runs, by human experts or other policies, or by decoding and exploration methods. This results in severe sample inefficiency and exploration difficulties, as well as a potential loss of diversity in the policy responses. Moreover, asynchronous PPO implementations require frequent and costly model transfers, and typically use value models which require a large amount of memory. In this paper we introduce Soft Policy Optimization (SPO), a simple, scalable and principled Soft RL method for sequence model policies that can learn from arbitrary online and offline trajectories and does not require a separate value model. In experiments on code contests, we shows that SPO outperforms PPO on pass@10, is significantly faster and more memory efficient, is able to benefit from off-policy data, enjoys improved stability, and learns more diverse (i.e. soft) policies.
☆ Linear-MoE: Linear Sequence Modeling Meets Mixture-of-Experts
Linear Sequence Modeling (LSM) like linear attention, state space models and linear RNNs, and Mixture-of-Experts (MoE) have recently emerged as significant architectural improvements. In this paper, we introduce Linear-MoE, a production-level system for modeling and training large-scale models that integrate LSM with MoE. Linear-MoE leverages the advantages of both LSM modules for linear-complexity sequence modeling and MoE layers for sparsely activation, aiming to offer high performance with efficient training. The Linear-MoE system comprises: 1) Modeling subsystem, which provides a unified framework supporting all instances of LSM. and 2) Training subsystem, which facilitates efficient training by incorporating various advanced parallelism technologies, particularly Sequence Parallelism designed for Linear-MoE models. Additionally, we explore hybrid models that combine Linear-MoE layers with standard Transformer-MoE layers with its Sequence Parallelism to further enhance model flexibility and performance. Evaluations on two model series, A0.3B-2B and A1B-7B, demonstrate Linear-MoE achieves efficiency gains while maintaining competitive performance on various benchmarks, showcasing its potential as a next-generation foundational model architecture. Code: https://github.com/OpenSparseLLMs/Linear-MoE.
comment: Technical report, 17 pages
☆ Quantum-PEFT: Ultra parameter-efficient fine-tuning ICLR 2025
This paper introduces Quantum-PEFT that leverages quantum computations for parameter-efficient fine-tuning (PEFT). Unlike other additive PEFT methods, such as low-rank adaptation (LoRA), Quantum-PEFT exploits an underlying full-rank yet surprisingly parameter efficient quantum unitary parameterization. With the use of Pauli parameterization, the number of trainable parameters grows only logarithmically with the ambient dimension, as opposed to linearly as in LoRA-based PEFT methods. Quantum-PEFT achieves vanishingly smaller number of trainable parameters than the lowest-rank LoRA as dimensions grow, enhancing parameter efficiency while maintaining a competitive performance. We apply Quantum-PEFT to several transfer learning benchmarks in language and vision, demonstrating significant advantages in parameter efficiency.
comment: ICLR 2025
☆ Towards Locally Explaining Prediction Behavior via Gradual Interventions and Measuring Property Gradients
Deep learning models achieve high predictive performance but lack intrinsic interpretability, hindering our understanding of the learned prediction behavior. Existing local explainability methods focus on associations, neglecting the causal drivers of model predictions. Other approaches adopt a causal perspective but primarily provide more general global explanations. However, for specific inputs, it's unclear whether globally identified factors apply locally. To address this limitation, we introduce a novel framework for local interventional explanations by leveraging recent advances in image-to-image editing models. Our approach performs gradual interventions on semantic properties to quantify the corresponding impact on a model's predictions using a novel score, the expected property gradient magnitude. We demonstrate the effectiveness of our approach through an extensive empirical evaluation on a wide range of architectures and tasks. First, we validate it in a synthetic scenario and demonstrate its ability to locally identify biases. Afterward, we apply our approach to analyze network training dynamics, investigate medical skin lesion classifiers, and study a pre-trained CLIP model with real-life interventional data. Our results highlight the potential of interventional explanations on the property level to reveal new insights into the behavior of deep models.
comment: 44 pages, 39 figures, 14 tables
☆ Semantic Shift Estimation via Dual-Projection and Classifier Reconstruction for Exemplar-Free Class-Incremental Learning
Exemplar-Free Class-Incremental Learning (EFCIL) aims to sequentially learn from distinct categories without retaining exemplars but easily suffers from catastrophic forgetting of learned knowledge. While existing EFCIL methods leverage knowledge distillation to alleviate forgetting, they still face two critical challenges: semantic shift and decision bias. Specifically, the embeddings of old tasks shift in the embedding space after learning new tasks, and the classifier becomes biased towards new tasks due to training solely with new data, thereby hindering the balance between old and new knowledge. To address these issues, we propose the Dual-Projection Shift Estimation and Classifier Reconstruction (DPCR) approach for EFCIL. DPCR effectively estimates semantic shift through a dual-projection, which combines a learnable transformation with a row-space projection to capture both task-wise and category-wise shifts. Furthermore, to mitigate decision bias, DPCR employs ridge regression to reformulate classifier training as a reconstruction process. This reconstruction exploits previous information encoded in covariance and prototype of each class after calibration with estimated shift, thereby reducing decision bias. Extensive experiments demonstrate that, across various datasets, DPCR effectively balances old and new tasks, outperforming state-of-the-art EFCIL methods.
comment: 14 pages, 7 figures
☆ Physics-based machine learning for fatigue lifetime prediction under non-uniform loading scenarios
Accurate lifetime prediction of structures subjected to cyclic loading is vital, especially in scenarios involving non-uniform loading histories where load sequencing critically influences structural durability. Addressing this complexity requires advanced modeling approaches capable of capturing the intricate relationship between loading sequences and fatigue lifetime. Traditional fatigue simulations are computationally prohibitive, necessitating more efficient methods. This study highlights the potential of physics-based machine learning ($\phi$ML) to predict the fatigue lifetime of materials. Specifically, a FFNN is designed to embed physical constraints from experimental evidence directly into its architecture to enhance prediction accuracy. It is trained using numerical simulations generated by a physically based anisotropic continuum damage fatigue model. The model is calibrated and validated against experimental fatigue data of concrete cylinder specimens tested in uniaxial compression. The proposed approach demonstrates superior accuracy compared to purely data-driven neural networks, particularly in situations with limited training data, achieving realistic predictions of damage accumulation. Thus, a general algorithm is developed and successfully applied to predict fatigue lifetimes under complex loading scenarios with multiple loading ranges. Hereby, the $\phi$ML model serves as a surrogate to capture damage evolution across load transitions. The $\phi$ML based algorithm is subsequently employed to investigate the influence of multiple loading transitions on accumulated fatigue life, and its predictions align with trends observed in recent experimental studies. This work demonstrates $\phi$ML as a promising technique for efficient and reliable fatigue life prediction in engineering structures, with possible integration into digital twin models for real-time assessment.
☆ R1-Omni: Explainable Omni-Multimodal Emotion Recognition with Reinforcing Learning
In this work, we present the first application of Reinforcement Learning with Verifiable Reward (RLVR) to an Omni-multimodal large language model in the context of emotion recognition, a task where both visual and audio modalities play crucial roles. We leverage RLVR to optimize the Omni model, significantly enhancing its performance in three key aspects: reasoning capability, emotion recognition accuracy, and generalization ability. The introduction of RLVR not only improves the model's overall performance on in-distribution data but also demonstrates superior robustness when evaluated on out-of-distribution datasets. More importantly, the improved reasoning capability enables clear analysis of the contributions of different modalities, particularly visual and audio information, in the emotion recognition process. This provides valuable insights into the optimization of multimodal large language models.
☆ Shifting Perspectives: Steering Vector Ensembles for Robust Bias Mitigation in LLMs ACL 2025
We present a novel approach to bias mitigation in large language models (LLMs) by applying steering vectors to modify model activations in forward passes. We employ Bayesian optimization to systematically identify effective contrastive pair datasets across nine bias axes. When optimized on the BBQ dataset, our individually tuned steering vectors achieve average improvements of 12.2%, 4.7%, and 3.2% over the baseline for Mistral, Llama, and Qwen, respectively. Building on these promising results, we introduce Steering Vector Ensembles (SVE), a method that averages multiple individually optimized steering vectors, each targeting a specific bias axis such as age, race, or gender. By leveraging their collective strength, SVE outperforms individual steering vectors in both bias reduction and maintaining model performance. The work presents the first systematic investigation of steering vectors for bias mitigation, and we demonstrate that SVE is a powerful and computationally efficient strategy for reducing bias in LLMs, with broader implications for enhancing AI safety.
comment: Submitted to ACL 2025
☆ Semi-Supervised Learning for Dose Prediction in Targeted Radionuclide: A Synthetic Data Study
Targeted Radionuclide Therapy (TRT) is a modern strategy in radiation oncology that aims to administer a potent radiation dose specifically to cancer cells using cancer-targeting radiopharmaceuticals. Accurate radiation dose estimation tailored to individual patients is crucial. Deep learning, particularly with pre-therapy imaging, holds promise for personalizing TRT doses. However, current methods require large time series of SPECT imaging, which is hardly achievable in routine clinical practice, and thus raises issues of data availability. Our objective is to develop a semi-supervised learning (SSL) solution to personalize dosimetry using pre-therapy images. The aim is to develop an approach that achieves accurate results when PET/CT images are available, but are associated with only a few post-therapy dosimetry data provided by SPECT images. In this work, we introduce an SSL method using a pseudo-label generation approach for regression tasks inspired by the FixMatch framework. The feasibility of the proposed solution was preliminarily evaluated through an in-silico study using synthetic data and Monte Carlo simulation. Experimental results for organ dose prediction yielded promising outcomes, showing that the use of pseudo-labeled data provides better accuracy compared to using only labeled data.
comment: 12 pages, 13 figures, 5 tables
☆ Improving Hate Speech Classification with Cross-Taxonomy Dataset Integration ACL
Algorithmic hate speech detection faces significant challenges due to the diverse definitions and datasets used in research and practice. Social media platforms, legal frameworks, and institutions each apply distinct yet overlapping definitions, complicating classification efforts. This study addresses these challenges by demonstrating that existing datasets and taxonomies can be integrated into a unified model, enhancing prediction performance and reducing reliance on multiple specialized classifiers. The work introduces a universal taxonomy and a hate speech classifier capable of detecting a wide range of definitions within a single framework. Our approach is validated by combining two widely used but differently annotated datasets, showing improved classification performance on an independent test set. This work highlights the potential of dataset and taxonomy integration in advancing hate speech detection, increasing efficiency, and ensuring broader applicability across contexts.
comment: Accepted for publication at LaTeCH-CLfL 2025. The 9th Joint ACL Special Interest Group on Language Technologies for the Socio-Economic Sciences and Humanities (SIGHUM) Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature
☆ Spatial Distillation based Distribution Alignment (SDDA) for Cross-Headset EEG Classification
A non-invasive brain-computer interface (BCI) enables direct interaction between the user and external devices, typically via electroencephalogram (EEG) signals. However, decoding EEG signals across different headsets remains a significant challenge due to differences in the number and locations of the electrodes. To address this challenge, we propose a spatial distillation based distribution alignment (SDDA) approach for heterogeneous cross-headset transfer in non-invasive BCIs. SDDA uses first spatial distillation to make use of the full set of electrodes, and then input/feature/output space distribution alignments to cope with the significant differences between the source and target domains. To our knowledge, this is the first work to use knowledge distillation in cross-headset transfers. Extensive experiments on six EEG datasets from two BCI paradigms demonstrated that SDDA achieved superior performance in both offline unsupervised domain adaptation and online supervised domain adaptation scenarios, consistently outperforming 10 classical and state-of-the-art transfer learning algorithms.
comment: 10 pages, 5 figures
☆ Toward an Evaluation Science for Generative AI Systems
There is an increasing imperative to anticipate and understand the performance and safety of generative AI systems in real-world deployment contexts. However, the current evaluation ecosystem is insufficient: Commonly used static benchmarks face validity challenges, and ad hoc case-by-case audits rarely scale. In this piece, we advocate for maturing an evaluation science for generative AI systems. While generative AI creates unique challenges for system safety engineering and measurement science, the field can draw valuable insights from the development of safety evaluation practices in other fields, including transportation, aerospace, and pharmaceutical engineering. In particular, we present three key lessons: Evaluation metrics must be applicable to real-world performance, metrics must be iteratively refined, and evaluation institutions and norms must be established. Applying these insights, we outline a concrete path toward a more rigorous approach for evaluating generative AI systems.
comment: First two authors contributed equally to this work
☆ Routing for Large ML Models
Training large language models (LLMs), and other large machine learning models, involves repeated communication of large volumes of data across a data center network. The communication patterns induced by these training process exhibit high regularity and persistence, giving rise to significant opportunities for optimizing the manner in which flows are routed across the network. We present an algorithmic framework for \textit{quantifying} network-wide efficiency in the context of training LLMs (and other large-scale ML models), and for periodically \textit{optimizing} routing with respect to this global metric.
☆ Graph Alignment via Birkhoff Relaxation
We consider the graph alignment problem, wherein the objective is to find a vertex correspondence between two graphs that maximizes the edge overlap. The graph alignment problem is an instance of the quadratic assignment problem (QAP), known to be NP-hard in the worst case even to approximately solve. In this paper, we analyze Birkhoff relaxation, a tight convex relaxation of QAP, and present theoretical guarantees on its performance when the inputs follow the Gaussian Wigner Model. More specifically, the weighted adjacency matrices are correlated Gaussian Orthogonal Ensemble with correlation $1/\sqrt{1+\sigma^2}$. Denote the optimal solutions of the QAP and Birkhoff relaxation by $\Pi^\star$ and $X^\star$ respectively. We show that $\|X^\star-\Pi^\star\|_F^2 = o(n)$ when $\sigma = o(n^{-1.25})$ and $\|X^\star-\Pi^\star\|_F^2 = \Omega(n)$ when $\sigma = \Omega(n^{-0.5})$. Thus, the optimal solution $X^\star$ transitions from a small perturbation of $\Pi^\star$ for small $\sigma$ to being well separated from $\Pi^\star$ as $\sigma$ becomes larger than $n^{-0.5}$. This result allows us to guarantee that simple rounding procedures on $X^\star$ align $1-o(1)$ fraction of vertices correctly whenever $\sigma = o(n^{-1.25})$. This condition on $\sigma$ to ensure the success of the Birkhoff relaxation is state-of-the-art.
☆ Riemannian Metric Learning: Closer to You than You Imagine
Riemannian metric learning is an emerging field in machine learning, unlocking new ways to encode complex data structures beyond traditional distance metric learning. While classical approaches rely on global distances in Euclidean space, they often fall short in capturing intrinsic data geometry. Enter Riemannian metric learning: a powerful generalization that leverages differential geometry to model the data according to their underlying Riemannian manifold. This approach has demonstrated remarkable success across diverse domains, from causal inference and optimal transport to generative modeling and representation learning. In this review, we bridge the gap between classical metric learning and Riemannian geometry, providing a structured and accessible overview of key methods, applications, and recent advances. We argue that Riemannian metric learning is not merely a technical refinement but a fundamental shift in how we think about data representations. Thus, this review should serve as a valuable resource for researchers and practitioners interested in exploring Riemannian metric learning and convince them that it is closer to them than they might imagine-both in theory and in practice.
☆ Disentangling Task Interference within Neurons: Model Merging in Alignment with Neuronal Mechanisms
Fine-tuning pre-trained models on targeted datasets enhances task-specific performance but often comes at the expense of generalization. Model merging techniques, which integrate multiple fine-tuned models into a single multi-task model through task arithmetic at various levels: model, layer, or parameter, offer a promising solution. However, task interference remains a fundamental challenge, leading to performance degradation and suboptimal merged models. Existing approaches largely overlook the fundamental role of individual neurons and their connectivity, resulting in a lack of interpretability in both the merging process and the merged models. In this work, we present the first study on the impact of neuronal alignment in model merging. We decompose task-specific representations into two complementary neuronal subspaces that regulate neuron sensitivity and input adaptability. Leveraging this decomposition, we introduce NeuroMerging, a novel merging framework developed to mitigate task interference within neuronal subspaces, enabling training-free model fusion across diverse tasks. Through extensive experiments, we demonstrate that NeuroMerging achieves superior performance compared to existing methods on multi-task benchmarks across both vision and natural language domains. Our findings highlight the importance of aligning neuronal mechanisms in model merging, offering new insights into mitigating task interference and improving knowledge fusion.
☆ Uncertainty-Aware Decoding with Minimum Bayes Risk ICLR 2025
Despite their outstanding performance in the majority of scenarios, contemporary language models still occasionally generate undesirable outputs, for example, hallucinated text. While such behaviors have previously been linked to uncertainty, there is a notable lack of methods that actively consider uncertainty during text generation. In this work, we show how Minimum Bayes Risk (MBR) decoding, which selects model generations according to an expected risk, can be generalized into a principled uncertainty-aware decoding method. In short, we account for model uncertainty during decoding by incorporating a posterior over model parameters into MBR's computation of expected risk. We show that this modified expected risk is useful for both choosing outputs and deciding when to abstain from generation and can provide improvements without incurring overhead. We benchmark different methods for learning posteriors and show that performance improves with prediction diversity. We release our code publicly.
comment: ICLR 2025 (Poster)
☆ CoinRobot: Generalized End-to-end Robotic Learning for Physical Intelligence
Physical intelligence holds immense promise for advancing embodied intelligence, enabling robots to acquire complex behaviors from demonstrations. However, achieving generalization and transfer across diverse robotic platforms and environments requires careful design of model architectures, training strategies, and data diversity. Meanwhile existing systems often struggle with scalability, adaptability to heterogeneous hardware, and objective evaluation in real-world settings. We present a generalized end-to-end robotic learning framework designed to bridge this gap. Our framework introduces a unified architecture that supports cross-platform adaptability, enabling seamless deployment across industrial-grade robots, collaborative arms, and novel embodiments without task-specific modifications. By integrating multi-task learning with streamlined network designs, it achieves more robust performance than conventional approaches, while maintaining compatibility with varying sensor configurations and action spaces. We validate our framework through extensive experiments on seven manipulation tasks. Notably, Diffusion-based models trained in our framework demonstrated superior performance and generalizability compared to the LeRobot framework, achieving performance improvements across diverse robotic platforms and environmental conditions.
☆ LoRACode: LoRA Adapters for Code Embeddings ICLR 2025
Code embeddings are essential for semantic code search; however, current approaches often struggle to capture the precise syntactic and contextual nuances inherent in code. Open-source models such as CodeBERT and UniXcoder exhibit limitations in scalability and efficiency, while high-performing proprietary systems impose substantial computational costs. We introduce a parameter-efficient fine-tuning method based on Low-Rank Adaptation (LoRA) to construct task-specific adapters for code retrieval. Our approach reduces the number of trainable parameters to less than two percent of the base model, enabling rapid fine-tuning on extensive code corpora (2 million samples in 25 minutes on two H100 GPUs). Experiments demonstrate an increase of up to 9.1% in Mean Reciprocal Rank (MRR) for Code2Code search, and up to 86.69% for Text2Code search tasks across multiple programming languages. Distinction in task-wise and language-wise adaptation helps explore the sensitivity of code retrieval for syntactical and linguistic variations.
comment: Accepted at the Deep Learning for Code (DL4C) Workshop at ICLR 2025
☆ Adversarial Policy Optimization for Offline Preference-based Reinforcement Learning
In this paper, we study offline preference-based reinforcement learning (PbRL), where learning is based on pre-collected preference feedback over pairs of trajectories. While offline PbRL has demonstrated remarkable empirical success, existing theoretical approaches face challenges in ensuring conservatism under uncertainty, requiring computationally intractable confidence set constructions. We address this limitation by proposing Adversarial Preference-based Policy Optimization (APPO), a computationally efficient algorithm for offline PbRL that guarantees sample complexity bounds without relying on explicit confidence sets. By framing PbRL as a two-player game between a policy and a model, our approach enforces conservatism in a tractable manner. Using standard assumptions on function approximation and bounded trajectory concentrability, we derive a sample complexity bound. To our knowledge, APPO is the first offline PbRL algorithm to offer both statistical efficiency and practical applicability. Experimental results on continuous control tasks demonstrate that APPO effectively learns from complex datasets, showing comparable performance with existing state-of-the-art methods.
☆ Robust Intrusion Detection System with Explainable Artificial Intelligence
Machine learning (ML) models serve as powerful tools for threat detection and mitigation; however, they also introduce potential new risks. Adversarial input can exploit these models through standard interfaces, thus creating new attack pathways that threaten critical network operations. As ML advancements progress, adversarial strategies become more advanced, and conventional defenses such as adversarial training are costly in computational terms and often fail to provide real-time detection. These methods typically require a balance between robustness and model performance, which presents challenges for applications that demand instant response. To further investigate this vulnerability, we suggest a novel strategy for detecting and mitigating adversarial attacks using eXplainable Artificial Intelligence (XAI). This approach is evaluated in real time within intrusion detection systems (IDS), leading to the development of a zero-touch mitigation strategy. Additionally, we explore various scenarios in the Radio Resource Control (RRC) layer within the Open Radio Access Network (O-RAN) framework, emphasizing the critical need for enhanced mitigation techniques to strengthen IDS defenses against advanced threats and implement a zero-touch mitigation solution. Extensive testing across different scenarios in the RRC layer of the O-RAN infrastructure validates the ability of the framework to detect and counteract integrated RRC-layer attacks when paired with adversarial strategies, emphasizing the essential need for robust defensive mechanisms to strengthen IDS against complex threats.
☆ An Analytical Model for Overparameterized Learning Under Class Imbalance
We study class-imbalanced linear classification in a high-dimensional Gaussian mixture model. We develop a tight, closed form approximation for the test error of several practical learning methods, including logit adjustment and class dependent temperature. Our approximation allows us to analytically tune and compare these methods, highlighting how and when they overcome the pitfalls of standard cross-entropy minimization. We test our theoretical findings on simulated data and imbalanced CIFAR10, MNIST and FashionMNIST datasets.
☆ Mastering Continual Reinforcement Learning through Fine-Grained Sparse Network Allocation and Dormant Neuron Exploration
Continual Reinforcement Learning (CRL) is essential for developing agents that can learn, adapt, and accumulate knowledge over time. However, a fundamental challenge persists as agents must strike a delicate balance between plasticity, which enables rapid skill acquisition, and stability, which ensures long-term knowledge retention while preventing catastrophic forgetting. In this paper, we introduce SSDE, a novel structure-based approach that enhances plasticity through a fine-grained allocation strategy with Structured Sparsity and Dormant-guided Exploration. SSDE decomposes the parameter space into forward-transfer (frozen) parameters and task-specific (trainable) parameters. Crucially, these parameters are allocated by an efficient co-allocation scheme under sparse coding, ensuring sufficient trainable capacity for new tasks while promoting efficient forward transfer through frozen parameters. However, structure-based methods often suffer from rigidity due to the accumulation of non-trainable parameters, limiting exploration and adaptability. To address this, we further introduce a sensitivity-guided neuron reactivation mechanism that systematically identifies and resets dormant neurons, which exhibit minimal influence in the sparse policy network during inference. This approach effectively enhance exploration while preserving structural efficiency. Extensive experiments on the CW10-v1 Continual World benchmark demonstrate that SSDE achieves state-of-the-art performance, reaching a success rate of 95%, surpassing prior methods significantly in both plasticity and stability trade-offs (code is available at: https://github.com/chengqiArchy/SSDE).
☆ Robust Conformal Prediction with a Single Binary Certificate ICLR 2025
Conformal prediction (CP) converts any model's output to prediction sets with a guarantee to cover the true label with (adjustable) high probability. Robust CP extends this guarantee to worst-case (adversarial) inputs. Existing baselines achieve robustness by bounding randomly smoothed conformity scores. In practice, they need expensive Monte-Carlo (MC) sampling (e.g. $\sim10^4$ samples per point) to maintain an acceptable set size. We propose a robust conformal prediction that produces smaller sets even with significantly lower MC samples (e.g. 150 for CIFAR10). Our approach binarizes samples with an adjustable (or automatically adjusted) threshold selected to preserve the coverage guarantee. Remarkably, we prove that robustness can be achieved by computing only one binary certificate, unlike previous methods that certify each calibration (or test) point. Thus, our method is faster and returns smaller robust sets. We also eliminate a previous limitation that requires a bounded score function.
comment: Published as a conference paper at ICLR 2025
☆ Guaranteeing Out-Of-Distribution Detection in Deep RL via Transition Estimation
An issue concerning the use of deep reinforcement learning (RL) agents is whether they can be trusted to perform reliably when deployed, as training environments may not reflect real-life environments. Anticipating instances outside their training scope, learning-enabled systems are often equipped with out-of-distribution (OOD) detectors that alert when a trained system encounters a state it does not recognize or in which it exhibits uncertainty. There exists limited work conducted on the problem of OOD detection within RL, with prior studies being unable to achieve a consensus on the definition of OOD execution within the context of RL. By framing our problem using a Markov Decision Process, we assume there is a transition distribution mapping each state-action pair to another state with some probability. Based on this, we consider the following definition of OOD execution within RL: A transition is OOD if its probability during real-life deployment differs from the transition distribution encountered during training. As such, we utilize conditional variational autoencoders (CVAE) to approximate the transition dynamics of the training environment and implement a conformity-based detector using reconstruction loss that is able to guarantee OOD detection with a pre-determined confidence level. We evaluate our detector by adapting existing benchmarks and compare it with existing OOD detection models for RL.
☆ Deep Sequence Models for Predicting Average Shear Wave Velocity from Strong Motion Records
This study explores the use of deep learning for predicting the time averaged shear wave velocity in the top 30 m of the subsurface ($V_{s30}$) at strong motion recording stations in T\"urkiye. $V_{s30}$ is a key parameter in site characterization and, as a result for seismic hazard assessment. However, it is often unavailable due to the lack of direct measurements and is therefore estimated using empirical correlations. Such correlations however are commonly inadequate in capturing complex, site-specific variability and this motivates the need for data-driven approaches. In this study, we employ a hybrid deep learning model combining convolutional neural networks (CNNs) and long short-term memory (LSTM) networks to capture both spatial and temporal dependencies in strong motion records. Furthermore, we explore how using different parts of the signal influence our deep learning model. Our results suggest that the hybrid approach effectively learns complex, nonlinear relationships within seismic signals. We observed that an improved P-wave arrival time model increased the prediction accuracy of $V_{s30}$. We believe the study provides valuable insights into improving $V_{s30}$ predictions using a CNN-LSTM framework, demonstrating its potential for improving site characterization for seismic studies. Our codes are available via this repo: https://github.com/brsylmz23/CNNLSTM_DeepEQ
☆ DiVISe: Direct Visual-Input Speech Synthesis Preserving Speaker Characteristics And Intelligibility NAACL 25
Video-to-speech (V2S) synthesis, the task of generating speech directly from silent video input, is inherently more challenging than other speech synthesis tasks due to the need to accurately reconstruct both speech content and speaker characteristics from visual cues alone. Recently, audio-visual pre-training has eliminated the need for additional acoustic hints in V2S, which previous methods often relied on to ensure training convergence. However, even with pre-training, existing methods continue to face challenges in achieving a balance between acoustic intelligibility and the preservation of speaker-specific characteristics. We analyzed this limitation and were motivated to introduce DiVISe (Direct Visual-Input Speech Synthesis), an end-to-end V2S model that predicts Mel-spectrograms directly from video frames alone. Despite not taking any acoustic hints, DiVISe effectively preserves speaker characteristics in the generated audio, and achieves superior performance on both objective and subjective metrics across the LRS2 and LRS3 datasets. Our results demonstrate that DiVISe not only outperforms existing V2S models in acoustic intelligibility but also scales more effectively with increased data and model parameters. Code and weights can be found at https://github.com/PussyCat0700/DiVISe.
comment: to be published in NAACL 25
☆ Robustness of Generalized Median Computation for Consensus Learning in Arbitrary Spaces
Robustness in terms of outliers is an important topic and has been formally studied for a variety of problems in machine learning and computer vision. Generalized median computation is a special instance of consensus learning and a common approach to finding prototypes. Related research can be found in numerous problem domains with a broad range of applications. So far, however, robustness of generalized median has only been studied in a few specific spaces. To our knowledge, there is no robustness characterization in a general setting, i.e. for arbitrary spaces. We address this open issue in our work. The breakdown point >=0.5 is proved for generalized median with metric distance functions in general. We also study the detailed behavior in case of outliers from different perspectives. In addition, we present robustness results for weighted generalized median computation and non-metric distance functions. Given the importance of robustness, our work contributes to closing a gap in the literature. The presented results have general impact and applicability, e.g. providing deeper understanding of generalized median computation and practical guidance to avoid non-robust computation.
☆ Policy Constraint by Only Support Constraint for Offline Reinforcement Learning
Offline reinforcement learning (RL) aims to optimize a policy by using pre-collected datasets, to maximize cumulative rewards. However, offline reinforcement learning suffers challenges due to the distributional shift between the learned and behavior policies, leading to errors when computing Q-values for out-of-distribution (OOD) actions. To mitigate this issue, policy constraint methods aim to constrain the learned policy's distribution with the distribution of the behavior policy or confine action selection within the support of the behavior policy. However, current policy constraint methods tend to exhibit excessive conservatism, hindering the policy from further surpassing the behavior policy's performance. In this work, we present Only Support Constraint (OSC) which is derived from maximizing the total probability of learned policy in the support of behavior policy, to address the conservatism of policy constraint. OSC presents a regularization term that only restricts policies to the support without imposing extra constraints on actions within the support. Additionally, to fully harness the performance of the new policy constraints, OSC utilizes a diffusion model to effectively characterize the support of behavior policies. Experimental evaluations across a variety of offline RL benchmarks demonstrate that OSC significantly enhances performance, alleviating the challenges associated with distributional shifts and mitigating conservatism of policy constraints. Code is available at https://github.com/MoreanP/OSC.
☆ Deep Muscle EMG construction using A Physics-Integrated Deep Learning approach
Electromyography (EMG)--based computational musculoskeletal modeling is a non-invasive method for studying musculotendon function, human movement, and neuromuscular control, providing estimates of internal variables like muscle forces and joint torques. However, EMG signals from deeper muscles are often challenging to measure by placing the surface EMG electrodes and unfeasible to measure directly using invasive methods. The restriction to the access of EMG data from deeper muscles poses a considerable obstacle to the broad adoption of EMG-driven modeling techniques. A strategic alternative is to use an estimation algorithm to approximate the missing EMG signals from deeper muscle. A similar strategy is used in physics-informed deep learning, where the features of physical systems are learned without labeled data. In this work, we propose a hybrid deep learning algorithm, namely the neural musculoskeletal model (NMM), that integrates physics-informed and data-driven deep learning to approximate the EMG signals from the deeper muscles. While data-driven modeling is used to predict the missing EMG signals, physics-based modeling engraves the subject-specific information into the predictions. Experimental verifications on five test subjects are carried out to investigate the performance of the proposed hybrid framework. The proposed NMM is validated against the joint torque computed from 'OpenSim' software. The predicted deep EMG signals are also compared against the state-of-the-art muscle synergy extrapolation (MSE) approach, where the proposed NMM completely outperforms the existing MSE framework by a significant margin.
☆ Uncertainty-Aware Explainable Federated Learning
Federated Learning (FL) is a collaborative machine learning paradigm for enhancing data privacy preservation. Its privacy-preserving nature complicates the explanation of the decision-making processes and the evaluation of the reliability of the generated explanations. In this paper, we propose the Uncertainty-aware eXplainable Federated Learning (UncertainXFL) to address these challenges. It generates explanations for decision-making processes under FL settings and provides information regarding the uncertainty of these explanations. UncertainXFL is the first framework to explicitly offer uncertainty evaluation for explanations within the FL context. Explanatory information is initially generated by the FL clients and then aggregated by the server in a comprehensive and conflict-free manner during FL training. The quality of the explanations, including the uncertainty score and tested validity, guides the FL training process by prioritizing clients with the most reliable explanations through higher weights during model aggregation. Extensive experimental evaluation results demonstrate that UncertainXFL achieves superior model accuracy and explanation accuracy, surpassing the current state-of-the-art model that does not incorporate uncertainty information by 2.71% and 1.77%, respectively. By integrating and quantifying uncertainty in the data into the explanation process, UncertainXFL not only clearly presents the explanation alongside its uncertainty, but also leverages this uncertainty to guide the FL training process, thereby enhancing the robustness and reliability of the resulting models.
☆ Safety-Critical Traffic Simulation with Adversarial Transfer of Driving Intentions ICRA 2025
Traffic simulation, complementing real-world data with a long-tail distribution, allows for effective evaluation and enhancement of the ability of autonomous vehicles to handle accident-prone scenarios. Simulating such safety-critical scenarios is nontrivial, however, from log data that are typically regular scenarios, especially in consideration of dynamic adversarial interactions between the future motions of autonomous vehicles and surrounding traffic participants. To address it, this paper proposes an innovative and efficient strategy, termed IntSim, that explicitly decouples the driving intentions of surrounding actors from their motion planning for realistic and efficient safety-critical simulation. We formulate the adversarial transfer of driving intention as an optimization problem, facilitating extensive exploration of diverse attack behaviors and efficient solution convergence. Simultaneously, intention-conditioned motion planning benefits from powerful deep models and large-scale real-world data, permitting the simulation of realistic motion behaviors for actors. Specially, through adapting driving intentions based on environments, IntSim facilitates the flexible realization of dynamic adversarial interactions with autonomous vehicles. Finally, extensive open-loop and closed-loop experiments on real-world datasets, including nuScenes and Waymo, demonstrate that the proposed IntSim achieves state-of-the-art performance in simulating realistic safety-critical scenarios and further improves planners in handling such scenarios.
comment: Accepted by ICRA 2025
☆ Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching
Recent advances in large language models have demonstrated remarkable reasoning capabilities through Chain of Thought (CoT) prompting, but often at the cost of excessive verbosity in their intermediate outputs, which increases computational overhead. We introduce Sketch-of-Thought (SoT), a novel prompting framework that combines cognitive-inspired reasoning paradigms with linguistic constraints to minimize token usage while preserving reasoning accuracy. SoT is designed as a flexible framework that can incorporate any custom reasoning paradigms based on cognitive science, and we instantiate it with three such paradigms - Conceptual Chaining, Chunked Symbolism, and Expert Lexicons - each tailored to different reasoning tasks and selected dynamically via a lightweight routing model. Through comprehensive evaluation across 15 reasoning datasets with multiple languages and multimodal scenarios, we demonstrate that SoT achieves token reductions of 76% with negligible accuracy impact. In certain domains like mathematical and multi-hop reasoning, it even improves accuracy while using significantly fewer tokens. Our code is publicly available: https://www.github.com/SimonAytes/SoT.
Self-Supervised Penalty-Based Learning for Robust Constrained Optimization
We propose a new methodology for parameterized constrained robust optimization, an important class of optimization problems under uncertainty, based on learning with a self-supervised penalty-based loss function. Whereas supervised learning requires pre-solved instances for training, our approach leverages a custom loss function derived from the exact penalty method in optimization to learn an approximation, typically defined by a neural network model, of the parameterized optimal solution mapping. Additionally, we adapt our approach to robust constrained combinatorial optimization problems by incorporating a surrogate linear cost over mixed integer domains, and a smooth approximations thereof, into the final layer of the network architecture. We perform computational experiments to test our approach on three different applications: multidimensional knapsack with continuous variables, combinatorial multidimensional knapsack with discrete variables, and an inventory management problem. Our results demonstrate that our self-supervised approach is able to effectively learn neural network approximations whose inference time is significantly smaller than the computation time of traditional solvers for this class of robust optimization problems. Furthermore, our results demonstrate that by varying the penalty parameter we are able to effectively balance the trade-off between sub-optimality and robust feasibility of the obtained solutions.
comment: To appear in the proceedings of CPAIOR 2025
☆ phepy: Visual Benchmarks and Improvements for Out-of-Distribution Detectors
Applying machine learning to increasingly high-dimensional problems with sparse or biased training data increases the risk that a model is used on inputs outside its training domain. For such out-of-distribution (OOD) inputs, the model can no longer make valid predictions, and its error is potentially unbounded. Testing OOD detection methods on real-world datasets is complicated by the ambiguity around which inputs are in-distribution (ID) or OOD. We design a benchmark for OOD detection, which includes three novel and easily-visualisable toy examples. These simple examples provide direct and intuitive insight into whether the detector is able to detect (1) linear and (2) non-linear concepts and (3) identify thin ID subspaces (needles) within high-dimensional spaces (haystacks). We use our benchmark to evaluate the performance of various methods from the literature. Since tactile examples of OOD inputs may benefit OOD detection, we also review several simple methods to synthesise OOD inputs for supervised training. We introduce two improvements, $t$-poking and OOD sample weighting, to make supervised detectors more precise at the ID-OOD boundary. This is especially important when conflicts between real ID and synthetic OOD sample blur the decision boundary. Finally, we provide recommendations for constructing and applying out-of-distribution detectors in machine learning.
☆ FMCHS: Advancing Traditional Chinese Medicine Herb Recommendation with Fusion of Multiscale Correlations of Herbs and Symptoms
Traditional Chinese medicine (TCM) exhibits remarkable therapeutic efficacy in disease treatment and healthcare through personalized herb prescriptions. However, current herb recommendation models inadequately capture the multiscale relations between herbs and clinical symptoms, particularly neglecting latent correlations at the chemical-molecular scale. To address these limitations, we propose the Fusion of Multiscale Correlations of Herbs and Symptoms (FMCHS), an innovative framework that synergistically integrates molecular-scale chemical characteristics of herbs with clinical symptoms. The framework employs multi-relational graph transformer layers to generate enriched embeddings that preserve both structural and semantic features within herbs and symptoms. Through systematic incorporation of herb chemical profiles into node embeddings and implementation of attention-based feature fusion, FMCHS effectively utilizes multiscale correlations. Comprehensive evaluations demonstrate FMCHS's superior performance over the state-of-the-art (SOTA) baseline, achieving relative improvements of 8.85% in Precision@5, 12.30% in Recall@5, and 10.86% in F1@5 compared to the SOTA model on benchmark datasets. This work facilitates the practical application of TCM in disease treatment and healthcare.
☆ Generative Trajectory Stitching through Diffusion Composition
Effective trajectory stitching for long-horizon planning is a significant challenge in robotic decision-making. While diffusion models have shown promise in planning, they are limited to solving tasks similar to those seen in their training data. We propose CompDiffuser, a novel generative approach that can solve new tasks by learning to compositionally stitch together shorter trajectory chunks from previously seen tasks. Our key insight is modeling the trajectory distribution by subdividing it into overlapping chunks and learning their conditional relationships through a single bidirectional diffusion model. This allows information to propagate between segments during generation, ensuring physically consistent connections. We conduct experiments on benchmark tasks of various difficulties, covering different environment sizes, agent state dimension, trajectory types, training data quality, and show that CompDiffuser significantly outperforms existing methods.
comment: Project page: https://comp-diffuser.github.io/
☆ Unity RL Playground: A Versatile Reinforcement Learning Framework for Mobile Robots
This paper introduces Unity RL Playground, an open-source reinforcement learning framework built on top of Unity ML-Agents. Unity RL Playground automates the process of training mobile robots to perform various locomotion tasks such as walking, running, and jumping in simulation, with the potential for seamless transfer to real hardware. Key features include one-click training for imported robot models, universal compatibility with diverse robot configurations, multi-mode motion learning capabilities, and extreme performance testing to aid in robot design optimization and morphological evolution. The attached video can be found at https://linqi-ye.github.io/video/iros25.mp4 and the code is coming soon.
☆ FedMABench: Benchmarking Mobile Agents on Decentralized Heterogeneous User Data
Mobile agents have attracted tremendous research participation recently. Traditional approaches to mobile agent training rely on centralized data collection, leading to high cost and limited scalability. Distributed training utilizing federated learning offers an alternative by harnessing real-world user data, providing scalability and reducing costs. However, pivotal challenges, including the absence of standardized benchmarks, hinder progress in this field. To tackle the challenges, we introduce FedMABench, the first benchmark for federated training and evaluation of mobile agents, specifically designed for heterogeneous scenarios. FedMABench features 6 datasets with 30+ subsets, 8 federated algorithms, 10+ base models, and over 800 apps across 5 categories, providing a comprehensive framework for evaluating mobile agents across diverse environments. Through extensive experiments, we uncover several key insights: federated algorithms consistently outperform local training; the distribution of specific apps plays a crucial role in heterogeneity; and, even apps from distinct categories can exhibit correlations during training. FedMABench is publicly available at: https://github.com/wwh0411/FedMABench with the datasets at: https://huggingface.co/datasets/wwh0411/FedMABench.
☆ Every FLOP Counts: Scaling a 300B Mixture-of-Experts LING LLM without Premium GPUs
In this technical report, we tackle the challenges of training large-scale Mixture of Experts (MoE) models, focusing on overcoming cost inefficiency and resource limitations prevalent in such systems. To address these issues, we present two differently sized MoE large language models (LLMs), namely Ling-Lite and Ling-Plus (referred to as "Bailing" in Chinese, spelled B\v{a}il\'ing in Pinyin). Ling-Lite contains 16.8 billion parameters with 2.75 billion activated parameters, while Ling-Plus boasts 290 billion parameters with 28.8 billion activated parameters. Both models exhibit comparable performance to leading industry benchmarks. This report offers actionable insights to improve the efficiency and accessibility of AI development in resource-constrained settings, promoting more scalable and sustainable technologies. Specifically, to reduce training costs for large-scale MoE models, we propose innovative methods for (1) optimization of model architecture and training processes, (2) refinement of training anomaly handling, and (3) enhancement of model evaluation efficiency. Additionally, leveraging high-quality data generated from knowledge graphs, our models demonstrate superior capabilities in tool use compared to other models. Ultimately, our experimental findings demonstrate that a 300B MoE LLM can be effectively trained on lower-performance devices while achieving comparable performance to models of a similar scale, including dense and MoE models. Compared to high-performance devices, utilizing a lower-specification hardware system during the pre-training phase demonstrates significant cost savings, reducing computing costs by approximately 20%. The models can be accessed at https://huggingface.co/inclusionAI.
comment: 34 pages
♻ ☆ Universality of Layer-Level Entropy-Weighted Quantization Beyond Model Architecture and Size
We present a novel approach to selective model quantization that transcends the limitations of architecture-specific and size-dependent compression methods for Large Language Models (LLMs) using Entropy-Weighted Quantization (EWQ). By analyzing the entropy distribution across transformer blocks, EWQ determines which blocks can be safely quantized without causing significant performance degradation, independent of model architecture or size. Our method outperforms uniform quantization approaches, maintaining Massive Multitask Language Understanding (MMLU) accuracy scores within 0.5% of unquantized models while reducing memory usage by up to 18%. We demonstrate the effectiveness of EWQ across multiple architectures -- from 1.6B to 70B parameters -- and showcase consistent improvements in the quality-compression trade-off regardless of model scale or architectural design. A surprising finding of EWQ is its ability to reduce perplexity compared to unquantized models, suggesting the presence of beneficial regularization through selective precision reduction. This improvement holds across different model families, indicating a fundamental relationship between layer-level entropy and optimal precision requirements. Additionally, we introduce FastEWQ, a rapid method for entropy distribution analysis that eliminates the need for loading model weights. This technique leverages universal characteristics of entropy distribution that persist across various architectures and scales, enabling near-instantaneous quantization decisions while maintaining 80% classification accuracy with full entropy analysis. Our results demonstrate that effective quantization strategies can be developed independently of specific architectural choices or model sizes, opening new possibilities for efficient LLM deployment.
comment: 29 pages, 7 figures, 14 tables; Fixed some types, added some clarifications and improvements
♻ ☆ No Forgetting Learning: Memory-free Continual Learning ICCV 2025
Continual Learning (CL) remains a central challenge in deep learning, where models must sequentially acquire new knowledge while mitigating Catastrophic Forgetting (CF) of prior tasks. Existing approaches often struggle with efficiency and scalability, requiring extensive memory or model buffers. This work introduces ``No Forgetting Learning" (NFL), a memory-free CL framework that leverages knowledge distillation to maintain stability while preserving plasticity. Memory-free means the NFL does not rely on any memory buffer. Through extensive evaluations of three benchmark datasets, we demonstrate that NFL achieves competitive performance while utilizing approximately 14.75 times less memory than state-of-the-art methods. Furthermore, we introduce a new metric to better assess CL's plasticity-stability trade-off.
comment: This paper is submitted to ICCV 2025
♻ ☆ Learning from negative feedback, or positive feedback or both
Existing preference optimization methods often assume scenarios where paired preference feedback (preferred/positive vs. dis-preferred/negative examples) is available. This requirement limits their applicability in scenarios where only unpaired feedback--for example, either positive or negative--is available. To address this, we introduce a novel approach that decouples learning from positive and negative feedback. This decoupling enables control over the influence of each feedback type and, importantly, allows learning even when only one feedback type is present. A key contribution is demonstrating stable learning from negative feedback alone, a capability not well-addressed by current methods. Our approach builds upon the probabilistic framework introduced in (Dayan and Hinton, 1997), which uses expectation-maximization (EM) to directly optimize the probability of positive outcomes (as opposed to classic expected reward maximization). We address a key limitation in current EM-based methods: they solely maximize the likelihood of positive examples, while neglecting negative ones. We show how to extend EM algorithms to explicitly incorporate negative examples, leading to a theoretically grounded algorithm that offers an intuitive and versatile way to learn from both positive and negative feedback. We evaluate our approach for training language models based on human feedback as well as training policies for sequential decision-making problems, where learned value functions are available.
♻ ☆ Speculative MoE: Communication Efficient Parallel MoE Inference with Speculative Token and Expert Pre-scheduling
MoE (Mixture of Experts) prevails as a neural architecture that can scale modern transformer-based LLMs (Large Language Models) to unprecedented scales. Nevertheless, large MoEs' great demands of computing power, memory capacity and memory bandwidth make scalable serving a fundamental challenge and efficient parallel inference has become a requisite to attain adequate throughput under latency constraints. DeepSpeed-MoE, one state-of-the-art MoE inference framework, adopts a 3D-parallel paradigm including EP (Expert Parallelism), TP (Tensor Parallel) and DP (Data Parallelism). However, our analysis shows DeepSpeed-MoE's inference efficiency is largely bottlenecked by EP, which is implemented with costly all-to-all collectives to route token activation. Our work aims to boost DeepSpeed-MoE by strategically reducing EP's communication overhead with a technique named Speculative MoE. Speculative MoE has two speculative parallelization schemes, speculative token shuffling and speculative expert grouping, which predict outstanding tokens' expert routing paths and pre-schedule tokens and experts across devices to losslessly trim EP's communication volume. Besides DeepSpeed-MoE, we also build Speculative MoE into a prevailing MoE inference engine SGLang. Experiments show Speculative MoE can significantly boost state-of-the-art MoE inference frameworks on fast homogeneous and slow heterogeneous interconnects.
♻ ☆ VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models
In this paper, we propose a novel framework for solving high-definition video inverse problems using latent image diffusion models. Building on recent advancements in spatio-temporal optimization for video inverse problems using image diffusion models, our approach leverages latent-space diffusion models to achieve enhanced video quality and resolution. To address the high computational demands of processing high-resolution frames, we introduce a pseudo-batch consistent sampling strategy, allowing efficient operation on a single GPU. Additionally, to improve temporal consistency, we present pseudo-batch inversion, an initialization technique that incorporates informative latents from the measurement. By integrating with SDXL, our framework achieves state-of-the-art video reconstruction across a wide range of spatio-temporal inverse problems, including complex combinations of frame averaging and various spatial degradations, such as deblurring, super-resolution, and inpainting. Unlike previous methods, our approach supports multiple aspect ratios (landscape, vertical, and square) and delivers HD-resolution reconstructions (exceeding 1280x720) in under 6 seconds per frame on a single NVIDIA 4090 GPU.
comment: Project page: https://vision-xl.github.io/
♻ ☆ Towards Autonomous Reinforcement Learning for Real-World Robotic Manipulation with Large Language Models
Recent advancements in Large Language Models (LLMs) and Visual Language Models (VLMs) have significantly impacted robotics, enabling high-level semantic motion planning applications. Reinforcement Learning (RL), a complementary paradigm, enables agents to autonomously optimize complex behaviors through interaction and reward signals. However, designing effective reward functions for RL remains challenging, especially in real-world tasks where sparse rewards are insufficient and dense rewards require elaborate design. In this work, we propose Autonomous Reinforcement learning for Complex HumanInformed Environments (ARCHIE), an unsupervised pipeline leveraging GPT-4, a pre-trained LLM, to generate reward functions directly from natural language task descriptions. The rewards are used to train RL agents in simulated environments, where we formalize the reward generation process to enhance feasibility. Additionally, GPT-4 automates the coding of task success criteria, creating a fully automated, one-shot procedure for translating human-readable text into deployable robot skills. Our approach is validated through extensive simulated experiments on single-arm and bi-manual manipulation tasks using an ABB YuMi collaborative robot, highlighting its practicality and effectiveness. Tasks are demonstrated on the real robot setup.
♻ ☆ Simple linear attention language models balance the recall-throughput tradeoff
Recent work has shown that attention-based language models excel at recall, the ability to ground generations in tokens previously seen in context. However, the efficiency of attention-based models is bottle-necked during inference by the KV-cache's aggressive memory consumption. In this work, we explore whether we can improve language model efficiency (e.g. by reducing memory consumption) without compromising on recall. By applying experiments and theory to a broad set of architectures, we identify a key tradeoff between a model's state size and recall ability. We show that efficient alternatives to attention (e.g. H3, Mamba, RWKV) maintain a fixed-size recurrent state, but struggle at recall. We propose BASED a simple architecture combining linear and sliding window attention. By varying BASED window size and linear attention feature dimension, we can dial the state size and traverse the pareto frontier of the recall-memory tradeoff curve, recovering the full quality of attention on one end and the small state size of attention-alternatives on the other. We train language models up to 1.3b parameters and show that BASED matches the strongest sub-quadratic models (e.g. Mamba) in perplexity and outperforms them on real-world recall-intensive tasks by 6.22 accuracy points. Implementations of linear attention are often less efficient than optimized standard attention implementations. To make BASED competitive, we develop IO-aware algorithms that enable 24x higher throughput on language generation than FlashAttention-2, when generating 1024 tokens using 1.3b parameter models. Code for this work is provided at: https://github.com/HazyResearch/based.
♻ ☆ Wasserstein Adaptive Value Estimation for Actor-Critic Reinforcement Learning
We present Wasserstein Adaptive Value Estimation for Actor-Critic (WAVE), an approach to enhance stability in deep reinforcement learning through adaptive Wasserstein regularization. Our method addresses the inherent instability of actor-critic algorithms by incorporating an adaptively weighted Wasserstein regularization term into the critic's loss function. We prove that WAVE achieves $\mathcal{O}\left(\frac{1}{k}\right)$ convergence rate for the critic's mean squared error and provide theoretical guarantees for stability through Wasserstein-based regularization. Using the Sinkhorn approximation for computational efficiency, our approach automatically adjusts the regularization based on the agent's performance. Theoretical analysis and experimental results demonstrate that WAVE achieves superior performance compared to standard actor-critic methods.
♻ ☆ Reliable and scalable variable importance estimation via warm-start and early stopping AISTATS
As opaque black-box predictive models become more prevalent, the need to develop interpretations for these models is of great interest. The concept of variable importance and Shapley values are interpretability measures that applies to any predictive model and assesses how much a variable or set of variables improves prediction performance. When the number of variables is large, estimating variable importance presents a significant computational challenge because re-training neural networks or other black-box algorithms requires significant additional computation. In this paper, we address this challenge for algorithms using gradient descent and gradient boosting (e.g. neural networks, gradient-boosted decision trees). By using the ideas of early stopping of gradient-based methods in combination with warm-start using the dropout method, we develop a scalable method to estimate variable importance for any algorithm that can be expressed as an iterative kernel update equation. Importantly, we provide theoretical guarantees by using the theory for early stopping of kernel-based methods for neural networks with sufficiently large (but not necessarily infinite) width and gradient-boosting decision trees that use symmetric trees as a weaker learner. We also demonstrate the efficacy of our methods through simulations and a real data example which illustrates the computational benefit of early stopping rather than fully re-training the model as well as the increased accuracy of our approach.
comment: Preliminary version accepted in AISTATS, 2025
♻ ☆ DeltaProduct: Increasing the Expressivity of DeltaNet Through Products of Householders ICLR 2025
Linear Recurrent Neural Networks (linear RNNs) have emerged as competitive alternatives to Transformers for sequence modeling, offering efficient training and linear-time inference. However, existing architectures face a fundamental trade-off between expressivity and efficiency, dictated by the structure of their state-transition matrices. While diagonal matrices used in architectures like Mamba, GLA, or mLSTM yield fast runtime, they suffer from severely limited expressivity. To address this, recent architectures such as (Gated) DeltaNet and RWKVv7 adopted a diagonal plus rank-1 structure, allowing simultaneous token-channel mixing, which overcomes some expressivity limitations with only a slight decrease in training efficiency. Building on the interpretation of DeltaNet's recurrence as performing one step of online gradient descent per token on an associative recall loss, we introduce DeltaProduct, which instead takes multiple ($n_h$) steps per token. This naturally leads to diagonal plus rank-$n_h$ state-transition matrices, formed as products of $n_h$ generalized Householder transformations, providing a tunable mechanism to balance expressivity and efficiency and a stable recurrence. Through extensive experiments, we demonstrate that DeltaProduct achieves superior state-tracking and language modeling capabilities while exhibiting significantly improved length extrapolation compared to DeltaNet. Additionally, we also strengthen the theoretical foundation of DeltaNet's expressivity by proving that it can solve dihedral group word problems in just two layers.
comment: Accepted at ICLR 2025 Workshop on Foundation Models in the Wild
♻ ☆ NeRF-Aug: Data Augmentation for Robotics with Neural Radiance Fields
Training a policy that can generalize to unknown objects is a long standing challenge within the field of robotics. The performance of a policy often drops significantly in situations where an object in the scene was not seen during training. To solve this problem, we present NeRF-Aug, a novel method that is capable of teaching a policy to interact with objects that are not present in the dataset. This approach differs from existing approaches by leveraging the speed, photorealism, and 3D consistency of a neural radiance field for augmentation. NeRF-Aug both creates more photorealistic data and runs 63% faster than existing methods. We demonstrate the effectiveness of our method on 5 tasks with 9 novel objects that are not present in the expert demonstrations. We achieve an average performance boost of 55.6% when comparing our method to the next best method. You can see video results at https://nerf-aug.github.io.
♻ ☆ Efficient Evolutionary Search Over Chemical Space with Large Language Models ICLR 2025
Molecular discovery, when formulated as an optimization problem, presents significant computational challenges because optimization objectives can be non-differentiable. Evolutionary Algorithms (EAs), often used to optimize black-box objectives in molecular discovery, traverse chemical space by performing random mutations and crossovers, leading to a large number of expensive objective evaluations. In this work, we ameliorate this shortcoming by incorporating chemistry-aware Large Language Models (LLMs) into EAs. Namely, we redesign crossover and mutation operations in EAs using LLMs trained on large corpora of chemical information. We perform extensive empirical studies on both commercial and open-source models on multiple tasks involving property optimization, molecular rediscovery, and structure-based drug design, demonstrating that the joint usage of LLMs with EAs yields superior performance over all baseline models across single- and multi-objective settings. We demonstrate that our algorithm improves both the quality of the final solution and convergence speed, thereby reducing the number of required objective evaluations. Our code is available at http://github.com/zoom-wang112358/MOLLEO
comment: Published in ICLR 2025
♻ ☆ Demystifying Misconceptions in Social Bots Research
Research on social bots aims at advancing knowledge and providing solutions to one of the most debated forms of online manipulation. Yet, social bot research is plagued by widespread biases, hyped results, and misconceptions that set the stage for ambiguities, unrealistic expectations, and seemingly irreconcilable findings. Overcoming such issues is instrumental towards ensuring reliable solutions and reaffirming the validity of the scientific method. In this contribution, we review some recent results in social bots research, highlighting and revising factual errors as well as methodological and conceptual biases. More importantly, we demystify common misconceptions, addressing fundamental points on how social bots research is discussed. Our analysis surfaces the need to discuss research about online disinformation and manipulation in a rigorous, unbiased, and responsible way. This article bolsters such effort by identifying and refuting common fallacious arguments used by both proponents and opponents of social bots research, as well as providing directions toward sound methodologies for future research in the field.
♻ ☆ TRADE: Transfer of Distributions between External Conditions with Normalizing Flows AISTATS 2025
Modeling distributions that depend on external control parameters is a common scenario in diverse applications like molecular simulations, where system properties like temperature affect molecular configurations. Despite the relevance of these applications, existing solutions are unsatisfactory as they require severely restricted model architectures or rely on energy-based training, which is prone to instability. We introduce TRADE, which overcomes these limitations by formulating the learning process as a boundary value problem. By initially training the model for a specific condition using either i.i.d.~samples or backward KL training, we establish a boundary distribution. We then propagate this information across other conditions using the gradient of the unnormalized density with respect to the external parameter. This formulation, akin to the principles of physics-informed neural networks, allows us to efficiently learn parameter-dependent distributions without restrictive assumptions. Experimentally, we demonstrate that TRADE achieves excellent results in a wide range of applications, ranging from Bayesian inference and molecular simulations to physical lattice models.
comment: Accepted as Poster at AISTATS 2025
♻ ☆ The NP-hardness of the Gromov-Wasserstein distance
This note addresses the property frequently mentioned in the literature that the Gromov-Wasserstein (GW) distance is NP-hard. We provide the details on the non-convex nature of the GW optimization problem that imply NP-hardness of the GW distance between finite spaces for any instance of an input data. We further illustrate the non-convexity of the problem with several explicit examples.
♻ ☆ BSAC-CoEx: Coexistence of URLLC and Distributed Learning Services via Device Selection
Recent advances in distributed intelligence have driven impressive progress across a diverse range of applications, from industrial automation to autonomous transportation. Nevertheless, deploying distributed learning services over wireless networks poses numerous challenges. These arise from inherent uncertainties in wireless environments (e.g., random channel fluctuations), limited resources (e.g., bandwidth and transmit power), and the presence of coexisting services on the network. In this paper, we investigate a mixed service scenario wherein high-priority ultra-reliable low latency communication (URLLC) and low-priority distributed learning services run concurrently over a network. Utilizing device selection, we aim to minimize the convergence time of distributed learning while simultaneously fulfilling the requirements of the URLLC service. We formulate this problem as a Markov decision process and address it via BSAC-CoEx, a framework based on the branching soft actor-critic (BSAC) algorithm that determines each device's participation decision through distinct branches in the actor's neural network. We evaluate our solution with a realistic simulator that is compliant with 3GPP standards for factory automation use cases. Our simulation results confirm that our solution can significantly decrease the training delays of the distributed learning service while keeping the URLLC availability above its required threshold and close to the scenario where URLLC solely consumes all wireless resources.
comment: This work has been submitted to the IEEE for possible publication
♻ ☆ SoK: Membership Inference Attacks on LLMs are Rushing Nowhere (and How to Fix It)
Whether LLMs memorize their training data and what this means, from measuring privacy leakage to detecting copyright violations, has become a rapidly growing area of research. In the last few months, more than 10 new methods have been proposed to perform Membership Inference Attacks (MIAs) against LLMs. Contrary to traditional MIAs which rely on fixed-but randomized-records or models, these methods are mostly trained and tested on datasets collected post-hoc. Sets of members and non-members, used to evaluate the MIA, are constructed using informed guesses after the release of a model. This lack of randomization raises concerns of a distribution shift between members and non-members. In this work, we first extensively review the literature on MIAs against LLMs and show that, while most work focuses on sequence-level MIAs evaluated in post-hoc setups, a range of target models, motivations and units of interest are considered. We then quantify distribution shifts present in 6 datasets used in the literature using a model-less bag of word classifier and show that all datasets constructed post-hoc suffer from strong distribution shifts. These shifts invalidate the claims of LLMs memorizing strongly in real-world scenarios and, potentially, also the methodological contributions of the recent papers based on these datasets. Yet, all hope might not be lost. We introduce important considerations to properly evaluate MIAs against LLMs and discuss, in turn, potential ways forwards: randomized test splits, injections of randomized (unique) sequences, randomized fine-tuning, and several post-hoc control methods. While each option comes with its advantages and limitations, we believe they collectively provide solid grounds to guide MIA development and study LLM memorization. We conclude with an overview of recommended approaches to benchmark sequence-level and document-level MIAs against LLMs.
comment: IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2025)
♻ ☆ Transformer-Based Fault-Tolerant Control for Fixed-Wing UAVs Using Knowledge Distillation and In-Context Adaptation
This study presents a transformer-based approach for fault-tolerant control in fixed-wing Unmanned Aerial Vehicles (UAVs), designed to adapt in real time to dynamic changes caused by structural damage or actuator failures. Unlike traditional Flight Control Systems (FCSs) that rely on classical control theory and struggle under severe alterations in dynamics, our method directly maps outer-loop reference values -- altitude, heading, and airspeed -- into control commands using the in-context learning and attention mechanisms of transformers, thus bypassing inner-loop controllers and fault-detection layers. Employing a teacher-student knowledge distillation framework, the proposed approach trains a student agent with partial observations by transferring knowledge from a privileged expert agent with full observability, enabling robust performance across diverse failure scenarios. Experimental results demonstrate that our transformer-based controller outperforms industry-standard FCS and state-of-the-art reinforcement learning (RL) methods, maintaining high tracking accuracy and stability in nominal conditions and extreme failure cases, highlighting its potential for enhancing UAV operational safety and reliability.
♻ ☆ Learning-Augmented Search Data Structures ICLR 2025
We study the integration of machine learning advice to improve upon traditional data structure designed for efficient search queries. Although there has been recent effort in improving the performance of binary search trees using machine learning advice, e.g., Lin et. al. (ICML 2022), the resulting constructions nevertheless suffer from inherent weaknesses of binary search trees, such as complexity of maintaining balance across multiple updates and the inability to handle partially-ordered or high-dimensional datasets. For these reasons, we focus on skip lists and KD trees in this work. Given access to a possibly erroneous oracle that outputs estimated fractional frequencies for search queries on a set of items, we construct skip lists and KD trees that provably provides the optimal expected search time, within nearly a factor of two. In fact, our learning-augmented skip lists and KD trees are still optimal up to a constant factor, even if the oracle is only accurate within a constant factor. We also demonstrate robustness by showing that our data structures achieves an expected search time that is within a constant factor of an oblivious skip list/KD tree construction even when the predictions are arbitrarily incorrect. Finally, we empirically show that our learning-augmented search data structures outperforms their corresponding traditional analogs on both synthetic and real-world datasets.
comment: ICLR 2025
♻ ☆ On the Completeness of Invariant Geometric Deep Learning Models
Invariant models, one important class of geometric deep learning models, are capable of generating meaningful geometric representations by leveraging informative geometric features in point clouds. These models are characterized by their simplicity, good experimental results and computational efficiency. However, their theoretical expressive power still remains unclear, restricting a deeper understanding of the potential of such models. In this work, we concentrate on characterizing the theoretical expressiveness of a wide range of invariant models under fully-connected conditions. We first rigorously characterize the expressiveness of the most classic invariant model, message-passing neural networks incorporating distance (DisGNN), restricting its unidentifiable cases to be only highly symmetric point clouds. We then prove that GeoNGNN, the geometric counterpart of one of the simplest subgraph graph neural networks, can effectively break these corner cases' symmetry and thus achieve E(3)-completeness. By leveraging GeoNGNN as a theoretical tool, we further prove that: 1) most subgraph GNNs developed in traditional graph learning can be seamlessly extended to geometric scenarios with E(3)-completeness; 2) DimeNet, GemNet and SphereNet, three well-established invariant models, are also all capable of achieving E(3)-completeness. Our theoretical results fill the gap in the expressive power of invariant models, contributing to a rigorous and comprehensive understanding of their capabilities.
comment: The Thirteenth International Conference on Learning Representations
♻ ☆ Bootstrapping Language Models with DPO Implicit Rewards ICLR 2025
Human alignment in large language models (LLMs) is an active area of research. A recent groundbreaking work, direct preference optimization (DPO), has greatly simplified the process from past work in reinforcement learning from human feedback (RLHF) by bypassing the reward learning stage in RLHF. DPO, after training, provides an implicit reward model. In this work, we make a novel observation that this implicit reward model can by itself be used in a bootstrapping fashion to further align the LLM. Our approach is to use the rewards from a current LLM to construct a preference dataset, which is then used in subsequent DPO rounds. We incorporate two refinements to further improve our approach: 1) length-regularized reward shaping to make the preference dataset length-unbiased; 2) experience replay to enhance the quality of the preference dataset. Our approach, named self-alignment with DPO ImpliCit rEwards (DICE), shows great improvements in alignment. It achieves an increase of more than 8$\\%$ in lengthcontrolled win rate on AlpacaEval 2 for all the different base models that we tried, without relying on external feedback. Our code is available at https://github.com/sail-sg/dice.
comment: Accepted in ICLR 2025
♻ ☆ Unsupervised detection of semantic correlations in big data
In real-world data, information is stored in extremely large feature vectors. These variables are typically correlated due to complex interactions involving many features simultaneously. Such correlations qualitatively correspond to semantic roles and are naturally recognized by both the human brain and artificial neural networks. This recognition enables, for instance, the prediction of missing parts of an image or text based on their context. We present a method to detect these correlations in high-dimensional data represented as binary numbers. We estimate the binary intrinsic dimension of a dataset, which quantifies the minimum number of independent coordinates needed to describe the data, and is therefore a proxy of semantic complexity. The proposed algorithm is largely insensitive to the so-called curse of dimensionality, and can therefore be used in big data analysis. We test this approach identifying phase transitions in model magnetic systems and we then apply it to the detection of semantic correlations of images and text inside deep neural networks.
♻ ☆ Massive Activations in Graph Neural Networks: Decoding Attention for Domain-Dependent Interpretability
Graph Neural Networks (GNNs) have become increasingly popular for effectively modeling graph-structured data, and attention mechanisms have been pivotal in enabling these models to capture complex patterns. In our study, we reveal a critical yet underexplored consequence of integrating attention into edge-featured GNNs: the emergence of Massive Activations (MAs) within attention layers. By developing a novel method for detecting MAs on edge features, we show that these extreme activations are not only activation anomalies but encode domain-relevant signals. Our post-hoc interpretability analysis demonstrates that, in molecular graphs, MAs aggregate predominantly on common bond types (e.g., single and double bonds) while sparing more informative ones (e.g., triple bonds). Furthermore, our ablation studies confirm that MAs can serve as natural attribution indicators, reallocating to less informative edges. Our study assesses various edge-featured attention-based GNN models using benchmark datasets, including ZINC, TOX21, and PROTEINS. Key contributions include (1) establishing the direct link between attention mechanisms and MAs generation in edge-featured GNNs, (2) developing a robust definition and detection method for MAs enabling reliable post-hoc interpretability. Overall, our study reveals the complex interplay between attention mechanisms, edge-featured GNNs model, and MAs emergence, providing crucial insights for relating GNNs internals to domain knowledge.
♻ ☆ Enhancing Architecture Frameworks by Including Modern Stakeholders and their Views/Viewpoints
Various architecture frameworks for software, systems, and enterprises have been proposed in the literature. They identified several stakeholders and defined modeling perspectives, architecture viewpoints, and views to frame and address stakeholder concerns. However, the stakeholders with data science and Machine Learning (ML) related concerns, such as data scientists and data engineers, are yet to be included in existing architecture frameworks. Only this way can we envision a holistic system architecture description of an ML-enabled system. Note that the ML component behavior and functionalities are special and should be distinguished from traditional software system behavior and functionalities. The main reason is that the actual functionality should be inferred from data instead of being specified at design time. Additionally, the structural models of ML components, such as ML model architectures, are typically specified using different notations and formalisms from what the Software Engineering (SE) community uses for software structural models. Yet, these two aspects, namely ML and non-ML, are becoming so intertwined that it necessitates an extension of software architecture frameworks and modeling practices toward supporting ML-enabled system architectures. In this paper, we address this gap through an empirical study using an online survey instrument. We surveyed 61 subject matter experts from over 25 organizations in 10 countries.
comment: ICICT 2025
♻ ☆ I/O in Machine Learning Applications on HPC Systems: A 360-degree Survey
Growing interest in Artificial Intelligence (AI) has resulted in a surge in demand for faster methods of Machine Learning (ML) model training and inference. This demand for speed has prompted the use of high performance computing (HPC) systems that excel in managing distributed workloads. Because data is the main fuel for AI applications, the performance of the storage and I/O subsystem of HPC systems is critical. In the past, HPC applications accessed large portions of data written by simulations or experiments or ingested data for visualizations or analysis tasks. ML workloads perform small reads spread across a large number of random files. This shift of I/O access patterns poses several challenges to modern parallel storage systems. In this paper, we survey I/O in ML applications on HPC systems, and target literature within a 6-year time window from 2019 to 2024. We define the scope of the survey, provide an overview of the common phases of ML, review available profilers and benchmarks, examine the I/O patterns encountered during offline data preparation, training, and inference, and explore I/O optimizations utilized in modern ML frameworks and proposed in recent literature. Lastly, we seek to expose research gaps that could spawn further R&D.
♻ ☆ CrystalGRW: Generative Modeling of Crystal Structures with Targeted Properties via Geodesic Random Walks
Determining whether a candidate crystalline material is thermodynamically stable depends on identifying its true ground-state structure, a central challenge in computational materials science. We introduce CrystalGRW, a diffusion-based generative model on Riemannian manifolds that proposes novel crystal configurations and can predict stable phases validated by density functional theory. The crystal properties, such as fractional coordinates, atomic types, and lattice matrices, are represented on suitable Riemannian manifolds, ensuring that new predictions generated through the diffusion process preserve the periodicity of crystal structures. We incorporate an equivariant graph neural network to also account for rotational and translational symmetries during the generation process. CrystalGRW demonstrates the ability to generate realistic crystal structures that are close to their ground states with accuracy comparable to existing models, while also enabling conditional control, such as specifying a desired crystallographic point group. These features help accelerate materials discovery and inverse design by offering stable, symmetry-consistent crystal candidates for experimental validation.
comment: 10+12 pages, 10 figures
♻ ☆ MeanCache: User-Centric Semantic Caching for LLM Web Services
Large Language Models (LLMs) like ChatGPT and Llama have revolutionized natural language processing and search engine dynamics. However, these models incur exceptionally high computational costs. For instance, GPT-3 consists of 175 billion parameters, where inference demands billions of floating-point operations. Caching is a natural solution to reduce LLM inference costs on repeated queries, which constitute about 31% of the total queries. However, existing caching methods are incapable of finding semantic similarities among LLM queries nor do they operate on contextual queries, leading to unacceptable false hit-and-miss rates. This paper introduces MeanCache, a user-centric semantic cache for LLM-based services that identifies semantically similar queries to determine cache hit or miss. Using MeanCache, the response to a user's semantically similar query can be retrieved from a local cache rather than re-querying the LLM, thus reducing costs, service provider load, and environmental impact. MeanCache leverages Federated Learning (FL) to collaboratively train a query similarity model without violating user privacy. By placing a local cache in each user's device and using FL, MeanCache reduces the latency and costs and enhances model performance, resulting in lower false hit rates. MeanCache also encodes context chains for every cached query, offering a simple yet highly effective mechanism to discern contextual query responses from standalone. Our experiments benchmarked against the state-of-the-art caching method, reveal that MeanCache attains an approximately 17% higher F-score and a 20% increase in precision during semantic cache hit-and-miss decisions while performing even better on contextual queries. It also reduces the storage requirement by 83% and accelerates semantic cache hit-and-miss decisions by 11%.
comment: Accepted at 2025 IEEE 39th International Parallel and Distributed Processing Symposium (IPDPS)
♻ ☆ Self-Learning for Personalized Keyword Spotting on Ultra-Low-Power Audio Sensors
This paper proposes a self-learning method to incrementally train (fine-tune) a personalized Keyword Spotting (KWS) model after the deployment on ultra-low power smart audio sensors. We address the fundamental problem of the absence of labeled training data by assigning pseudo-labels to the new recorded audio frames based on a similarity score with respect to few user recordings. By experimenting with multiple KWS models with a number of parameters up to 0.5M on two public datasets, we show an accuracy improvement of up to +19.2% and +16.0% vs. the initial models pretrained on a large set of generic keywords. The labeling task is demonstrated on a sensor system composed of a low-power microphone and an energy-efficient Microcontroller (MCU). By efficiently exploiting the heterogeneous processing engines of the MCU, the always-on labeling task runs in real-time with an average power cost of up to 8.2 mW. On the same platform, we estimate an energy cost for on-device training 10x lower than the labeling energy if sampling a new utterance every 6.1 s or 18.8 s with a DS-CNN-S or a DS-CNN-M model. Our empirical result paves the way to self-adaptive personalized KWS sensors at the extreme edge.
comment: Published on IEEE IoT Journal
♻ ☆ Equivariant Neural Functional Networks for Transformers ICLR 2025
This paper systematically explores neural functional networks (NFN) for transformer architectures. NFN are specialized neural networks that treat the weights, gradients, or sparsity patterns of a deep neural network (DNN) as input data and have proven valuable for tasks such as learnable optimizers, implicit data representations, and weight editing. While NFN have been extensively developed for MLP and CNN, no prior work has addressed their design for transformers, despite the importance of transformers in modern deep learning. This paper aims to address this gap by providing a systematic study of NFN for transformers. We first determine the maximal symmetric group of the weights in a multi-head attention module as well as a necessary and sufficient condition under which two sets of hyperparameters of the multi-head attention module define the same function. We then define the weight space of transformer architectures and its associated group action, which leads to the design principles for NFN in transformers. Based on these, we introduce Transformer-NFN, an NFN that is equivariant under this group action. Additionally, we release a dataset of more than 125,000 Transformers model checkpoints trained on two datasets with two different tasks, providing a benchmark for evaluating Transformer-NFN and encouraging further research on transformer training and performance.
comment: Accepted in ICLR 2025
♻ ☆ A Novel Hybrid Approach to Contraceptive Demand Forecasting: Integrating Point Predictions with Probabilistic Distributions
Accurate demand forecasting is vital for ensuring reliable access to contraceptive products, supporting key processes like procurement, inventory, and distribution. However, forecasting contraceptive demand in developing countries presents challenges, including incomplete data, poor data quality, and the need to account for multiple geographical and product factors. Current methods often rely on simple forecasting techniques, which fail to capture demand uncertainties arising from these factors, warranting expert involvement. Our study aims to improve contraceptive demand forecasting by combining probabilistic forecasting methods with expert knowledge. We developed a hybrid model that combines point forecasts from domain-specific model with probabilistic distributions from statistical and machine learning approaches, enabling human input to fine-tune and enhance the system-generated forecasts. This approach helps address the uncertainties in demand and is particularly useful in resource-limited settings. We evaluate different forecasting methods, including time series, Bayesian, machine learning, and foundational time series methods alongside our new hybrid approach. By comparing these methods, we provide insights into their strengths, weaknesses, and computational requirements. Our research fills a gap in forecasting contraceptive demand and offers a practical framework that combines algorithmic and human expertise. Our proposed model can also be generalized to other humanitarian contexts with similar data patterns.
♻ ☆ Cauchy-Schwarz Regularizers ICLR 2025
We introduce a novel class of regularization functions, called Cauchy-Schwarz (CS) regularizers, which can be designed to induce a wide range of properties in solution vectors of optimization problems. To demonstrate the versatility of CS regularizers, we derive regularization functions that promote discrete-valued vectors, eigenvectors of a given matrix, and orthogonal matrices. The resulting CS regularizers are simple, differentiable, and can be free of spurious stationary points, making them suitable for gradient-based solvers and large-scale optimization problems. In addition, CS regularizers automatically adapt to the appropriate scale, which is, for example, beneficial when discretizing the weights of neural networks. To demonstrate the efficacy of CS regularizers, we provide results for solving underdetermined systems of linear equations and weight quantization in neural networks. Furthermore, we discuss specializations, variations, and generalizations, which lead to an even broader class of new and possibly more powerful regularizers.
comment: Accepted to ICLR 2025
♻ ☆ OASIS Uncovers: High-Quality T2I Models, Same Old Stereotypes ICLR 2025
Images generated by text-to-image (T2I) models often exhibit visual biases and stereotypes of concepts such as culture and profession. Existing quantitative measures of stereotypes are based on statistical parity that does not align with the sociological definition of stereotypes and, therefore, incorrectly categorizes biases as stereotypes. Instead of oversimplifying stereotypes as biases, we propose a quantitative measure of stereotypes that aligns with its sociological definition. We then propose OASIS to measure the stereotypes in a generated dataset and understand their origins within the T2I model. OASIS includes two scores to measure stereotypes from a generated image dataset: (M1) Stereotype Score to measure the distributional violation of stereotypical attributes, and (M2) WALS to measure spectral variance in the images along a stereotypical attribute. OASIS also includes two methods to understand the origins of stereotypes in T2I models: (U1) StOP to discover attributes that the T2I model internally associates with a given concept, and (U2) SPI to quantify the emergence of stereotypical attributes in the latent space of the T2I model during image generation. Despite the considerable progress in image fidelity, using OASIS, we conclude that newer T2I models such as FLUX.1 and SDv3 contain strong stereotypical predispositions about concepts and still generate images with widespread stereotypical attributes. Additionally, the quantity of stereotypes worsens for nationalities with lower Internet footprints.
comment: Accepted as a Spotlight paper at ICLR 2025
♻ ☆ AI, Meet Human: Learning Paradigms for Hybrid Decision Making Systems
Everyday we increasingly rely on machine learning models to automate and support high-stake tasks and decisions. This growing presence means that humans are now constantly interacting with machine learning-based systems, training and using models everyday. Several different techniques in computer science literature account for the human interaction with machine learning systems, but their classification is sparse and the goals varied. This survey proposes a taxonomy of Hybrid Decision Making Systems, providing both a conceptual and technical framework for understanding how current computer science literature models interaction between humans and machines.
♻ ☆ Membership Inference Attacks Cannot Prove that a Model Was Trained On Your Data
We consider the problem of a training data proof, where a data creator or owner wants to demonstrate to a third party that some machine learning model was trained on their data. Training data proofs play a key role in recent lawsuits against foundation models trained on web-scale data. Many prior works suggest to instantiate training data proofs using membership inference attacks. We argue that this approach is fundamentally unsound: to provide convincing evidence, the data creator needs to demonstrate that their attack has a low false positive rate, i.e., that the attack's output is unlikely under the null hypothesis that the model was not trained on the target data. Yet, sampling from this null hypothesis is impossible, as we do not know the exact contents of the training set, nor can we (efficiently) retrain a large foundation model. We conclude by offering two paths forward, by showing that data extraction attacks and membership inference on special canary data can be used to create sound training data proofs.
comment: position paper at IEEE SaTML 2025
♻ ☆ Demystifying the Token Dynamics of Deep Selective State Space Models ICLR 2025
Selective state space models (SSM), such as Mamba, have gained prominence for their effectiveness in modeling sequential data. Despite their outstanding empirical performance, a comprehensive theoretical understanding of deep selective SSM remains elusive, hindering their further development and adoption for applications that need high fidelity. In this paper, we investigate the dynamical properties of tokens in a pre-trained Mamba model. In particular, we derive the dynamical system governing the continuous-time limit of the Mamba model and characterize the asymptotic behavior of its solutions. In the one-dimensional case, we prove that only one of the following two scenarios happens: either all tokens converge to zero, or all tokens diverge to infinity. We provide criteria based on model parameters to determine when each scenario occurs. For the convergent scenario, we empirically verify that this scenario negatively impacts the model's performance. For the divergent scenario, we prove that different tokens will diverge to infinity at different rates, thereby contributing unequally to the updates during model training. Based on these investigations, we propose two refinements for the model: excluding the convergent scenario and reordering tokens based on their importance scores, both aimed at improving practical performance. Our experimental results validate these refinements, offering insights into enhancing Mamba's effectiveness in real-world applications.
comment: Accepted at ICLR 2025 (spotlight)
♻ ☆ Toward Robust Non-Transferable Learning: A Survey and Benchmark
Over the past decades, researchers have primarily focused on improving the generalization abilities of models, with limited attention given to regulating such generalization. However, the ability of models to generalize to unintended data (e.g., harmful or unauthorized data) can be exploited by malicious adversaries in unforeseen ways, potentially resulting in violations of model ethics. Non-transferable learning (NTL), a task aimed at reshaping the generalization abilities of deep learning models, was proposed to address these challenges. While numerous methods have been proposed in this field, a comprehensive review of existing progress and a thorough analysis of current limitations remain lacking. In this paper, we bridge this gap by presenting the first comprehensive survey on NTL and introducing NTLBench, the first benchmark to evaluate NTL performance and robustness within a unified framework. Specifically, we first introduce the task settings, general framework, and criteria of NTL, followed by a summary of NTL approaches. Furthermore, we emphasize the often-overlooked issue of robustness against various attacks that can destroy the non-transferable mechanism established by NTL. Experiments conducted via NTLBench verify the limitations of existing NTL methods in robustness. Finally, we discuss the practical applications of NTL, along with its future directions and associated challenges.
comment: Code is available at https://github.com/tmllab/NTLBench
♻ ☆ Stochastic Modified Flows for Riemannian Stochastic Gradient Descent
We give quantitative estimates for the rate of convergence of Riemannian stochastic gradient descent (RSGD) to Riemannian gradient flow and to a diffusion process, the so-called Riemannian stochastic modified flow (RSMF). Using tools from stochastic differential geometry we show that, in the small learning rate regime, RSGD can be approximated by the solution to the RSMF driven by an infinite-dimensional Wiener process. The RSMF accounts for the random fluctuations of RSGD and, thereby, increases the order of approximation compared to the deterministic Riemannian gradient flow. The RSGD is build using the concept of a retraction map, that is, a cost efficient approximation of the exponential map, and we prove quantitative bounds for the weak error of the diffusion approximation under assumptions on the retraction map, the geometry of the manifold, and the random estimators of the gradient.
♻ ☆ Analysis of the BraTS 2023 Intracranial Meningioma Segmentation Challenge MICCAI
We describe the design and results from the BraTS 2023 Intracranial Meningioma Segmentation Challenge. The BraTS Meningioma Challenge differed from prior BraTS Glioma challenges in that it focused on meningiomas, which are typically benign extra-axial tumors with diverse radiologic and anatomical presentation and a propensity for multiplicity. Nine participating teams each developed deep-learning automated segmentation models using image data from the largest multi-institutional systematically expert annotated multilabel multi-sequence meningioma MRI dataset to date, which included 1000 training set cases, 141 validation set cases, and 283 hidden test set cases. Each case included T2, FLAIR, T1, and T1Gd brain MRI sequences with associated tumor compartment labels delineating enhancing tumor, non-enhancing tumor, and surrounding non-enhancing FLAIR hyperintensity. Participant automated segmentation models were evaluated and ranked based on a scoring system evaluating lesion-wise metrics including dice similarity coefficient (DSC) and 95% Hausdorff Distance. The top ranked team had a lesion-wise median dice similarity coefficient (DSC) of 0.976, 0.976, and 0.964 for enhancing tumor, tumor core, and whole tumor, respectively and a corresponding average DSC of 0.899, 0.904, and 0.871, respectively. These results serve as state-of-the-art benchmarks for future pre-operative meningioma automated segmentation algorithms. Additionally, we found that 1286 of 1424 cases (90.3%) had at least 1 compartment voxel abutting the edge of the skull-stripped image edge, which requires further investigation into optimal pre-processing face anonymization steps.
comment: Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA) https://melba-journal.org/2025:003 22 pages, 6 tables, 12 figures, MICCAI, MELBA
♻ ☆ Hints-In-Browser: Benchmarking Language Models for Programming Feedback Generation
Generative AI and large language models hold great promise in enhancing programming education by generating individualized feedback and hints for learners. Recent works have primarily focused on improving the quality of generated feedback to achieve human tutors' quality. While quality is an important performance criterion, it is not the only criterion to optimize for real-world educational deployments. In this paper, we benchmark language models for programming feedback generation across several performance criteria, including quality, cost, time, and data privacy. The key idea is to leverage recent advances in the new paradigm of in-browser inference that allow running these models directly in the browser, thereby providing direct benefits across cost and data privacy. To boost the feedback quality of small models compatible with in-browser inference engines, we develop a fine-tuning pipeline based on GPT-4 generated synthetic data. We showcase the efficacy of fine-tuned Llama3-8B and Phi3-3.8B 4-bit quantized models using WebLLM's in-browser inference engine on three different Python programming datasets. We will release the full implementation along with a web app and datasets to facilitate further research on in-browser language models.
♻ ☆ Sustainable transparency in Recommender Systems: Bayesian Ranking of Images for Explainability
Recommender Systems have become crucial in the modern world, commonly guiding users towards relevant content or products, and having a large influence over the decisions of users and citizens. However, ensuring transparency and user trust in these systems remains a challenge; personalized explanations have emerged as a solution, offering justifications for recommendations. Among the existing approaches for generating personalized explanations, using existing visual content created by users is a promising option to maximize transparency and user trust. State-of-the-art models that follow this approach, despite leveraging highly optimized architectures, employ surrogate learning tasks that do not efficiently model the objective of ranking images as explanations for a given recommendation; this leads to a suboptimal training process with high computational costs that may not be reduced without affecting model performance. This work presents BRIE, a novel model where we leverage Bayesian Pairwise Ranking to enhance the training process, allowing us to consistently outperform state-of-the-art models in six real-world datasets while reducing its model size by up to 64 times and its CO2 emissions by up to 75% in training and inference.
♻ ☆ RULSurv: A probabilistic survival-based method for early censoring-aware prediction of remaining useful life in ball bearings
Censored data refers to situations where the full information about a particular event or process is only partially known. In survival analysis, censoring plays an important role, as ignoring such observations can bias the model parameters and overestimate the probability of when the event is likely to occur. There has been a renewed interest in using data-driven methods to predict the remaining useful life (RUL) of ball bearings for predictive maintenance. However, few studies have explicitly addressed the challenge of handling censored data. To address this issue, we introduce a novel and flexible method for early fault detection using Kullback-Leibler (KL) divergence and RUL estimation using survival analysis that naturally supports censored data. We demonstrate our approach in the XJTU-SY dataset using a 5-fold cross-validation across three different operating conditions. When predicting the time to failure for bearings under the highest load (C1, 12.0 kN and 2100 RPM) with 25\% random censoring, our approach achieves a mean absolute error (MAE) of 14.7 minutes (95\% CI 13.6-15.8) using a linear CoxPH model, and an MAE of 12.6 minutes (95\% CI 11.8-13.4) using a nonlinear Random Survival Forests model, compared to an MAE of 18.5 minutes (95\% 17.4-19.6) using a linear LASSO model that does not support censoring. Moreover, our approach achieves a mean cumulative relative accuracy (CRA) of 0.7586 over 5 bearings under the highest load, which improves over several state-of-the-art baselines. Our work highlights the importance of considering censored observations as part of the model design when building predictive models for early fault detection and RUL estimation.
♻ ☆ Positive-Unlabelled Learning for identifying new candidate Dietary Restriction-related genes among Ageing-related genes
Dietary Restriction (DR) is one of the most popular anti-ageing interventions; recently, Machine Learning (ML) has been explored to identify potential DR-related genes among ageing-related genes, aiming to minimize costly wet lab experiments needed to expand our knowledge on DR. However, to train a model from positive (DR-related) and negative (non-DR-related) examples, the existing ML approach naively labels genes without known DR relation as negative examples, assuming that lack of DR-related annotation for a gene represents evidence of absence of DR-relatedness, rather than absence of evidence. This hinders the reliability of the negative examples (non-DR-related genes) and the method's ability to identify novel DR-related genes. This work introduces a novel gene prioritisation method based on the two-step Positive-Unlabelled (PU) Learning paradigm: using a similarity-based, KNN-inspired approach, our method first selects reliable negative examples among the genes without known DR associations. Then, these reliable negatives and all known positives are used to train a classifier that effectively differentiates DR-related and non-DR-related genes, which is finally employed to generate a more reliable ranking of promising genes for novel DR-relatedness. Our method significantly outperforms (p<0.05) the existing state-of-the-art approach in three predictive accuracy metrics with up to 40% lower computational cost in the best case, and we identify 4 new promising DR-related genes (PRKAB1, PRKAB2, IRS2, PRKAG1), all with evidence from the existing literature supporting their potential DR-related role.
♻ ☆ DeFoG: Discrete Flow Matching for Graph Generation
Graph generative models are essential across diverse scientific domains by capturing complex distributions over relational data. Among them, graph diffusion models achieve superior performance but face inefficient sampling and limited flexibility due to the tight coupling between training and sampling stages. We introduce DeFoG, a novel graph generative framework that disentangles sampling from training, enabling a broader design space for more effective and efficient model optimization. DeFoG employs a discrete flow-matching formulation that respects the inherent symmetries of graphs. We theoretically ground this disentangled formulation by explicitly relating the training loss to the sampling algorithm and showing that DeFoG faithfully replicates the ground truth graph distribution. Building on these foundations, we thoroughly investigate DeFoG's design space and propose novel sampling methods that significantly enhance performance and reduce the required number of refinement steps. Extensive experiments demonstrate state-of-the-art performance across synthetic, molecular, and digital pathology datasets, covering both unconditional and conditional generation settings. It also outperforms most diffusion-based models with just 5-10% of their sampling steps.
♻ ☆ On Forecast Stability
Forecasts are typically not produced in a vacuum but in a business context, where forecasts are generated on a regular basis and interact with each other. For decisions, it may be important that forecasts do not change arbitrarily, and are stable in some sense. However, this area has received only limited attention in the forecasting literature. In this paper, we explore two types of forecast stability that we call vertical stability and horizontal stability. The existing works in the literature are only applicable to certain base models and extending these frameworks to be compatible with any base model is not straightforward. Furthermore, these frameworks can only stabilise the forecasts vertically. To fill this gap, we propose a simple linear-interpolation-based approach that is applicable to stabilise the forecasts provided by any base model vertically and horizontally. The approach can produce both accurate and stable forecasts. Using N-BEATS, Pooled Regression and LightGBM as the base models, in our evaluation on four publicly available datasets, the proposed framework is able to achieve significantly higher stability and/or accuracy compared to a set of benchmarks including a state-of-the-art forecast stabilisation method across three error metrics and six stability metrics.
♻ ☆ A Confidence-based Acquisition Model for Self-supervised Active Learning and Label Correction
Supervised neural approaches are hindered by their dependence on large, meticulously annotated datasets, a requirement that is particularly cumbersome for sequential tasks. The quality of annotations tends to deteriorate with the transition from expert-based to crowd-sourced labelling. To address these challenges, we present CAMEL (Confidence-based Acquisition Model for Efficient self-supervised active Learning), a pool-based active learning framework tailored to sequential multi-output problems. CAMEL possesses two core features: (1) it requires expert annotators to label only a fraction of a chosen sequence, and (2) it facilitates self-supervision for the remainder of the sequence. By deploying a label correction mechanism, CAMEL can also be utilised for data cleaning. We evaluate CAMEL on two sequential tasks, with a special emphasis on dialogue belief tracking, a task plagued by the constraints of limited and noisy datasets. Our experiments demonstrate that CAMEL significantly outperforms the baselines in terms of efficiency. Furthermore, the data corrections suggested by our method contribute to an overall improvement in the quality of the resulting datasets.
♻ ☆ Offline Safe Reinforcement Learning Using Trajectory Classification AAAI 2025
Offline safe reinforcement learning (RL) has emerged as a promising approach for learning safe behaviors without engaging in risky online interactions with the environment. Most existing methods in offline safe RL rely on cost constraints at each time step (derived from global cost constraints) and this can result in either overly conservative policies or violation of safety constraints. In this paper, we propose to learn a policy that generates desirable trajectories and avoids undesirable trajectories. To be specific, we first partition the pre-collected dataset of state-action trajectories into desirable and undesirable subsets. Intuitively, the desirable set contains high reward and safe trajectories, and undesirable set contains unsafe trajectories and low-reward safe trajectories. Second, we learn a policy that generates desirable trajectories and avoids undesirable trajectories, where (un)desirability scores are provided by a classifier learnt from the dataset of desirable and undesirable trajectories. This approach bypasses the computational complexity and stability issues of a min-max objective that is employed in existing methods. Theoretically, we also show our approach's strong connections to existing learning paradigms involving human feedback. Finally, we extensively evaluate our method using the DSRL benchmark for offline safe RL. Empirically, our method outperforms competitive baselines, achieving higher rewards and better constraint satisfaction across a wide variety of benchmark tasks.
comment: AAAI 2025
♻ ☆ EdgeMoE: Empowering Sparse Large Language Models on Mobile Devices
Large language models (LLMs) such as GPTs and Mixtral-8x7B have revolutionized machine intelligence due to their exceptional abilities in generic ML tasks. Transiting LLMs from datacenters to edge devices brings benefits like better privacy and availability, but is challenged by their massive parameter size and thus unbearable runtime costs. To this end, we present EdgeMoE, an on-device inference engine for mixture-of-expert (MoE) LLMs -- a popular form of sparse LLM that scales its parameter size with almost constant computing complexity. EdgeMoE achieves both memory- and compute-efficiency by partitioning the model into the storage hierarchy: non-expert weights are held in device memory; while expert weights are held on external storage and fetched to memory only when activated. This design is motivated by a key observation that expert weights are bulky but infrequently used due to sparse activation. To further reduce the expert I/O swapping overhead, EdgeMoE incorporates two novel techniques: (1) expert-wise bitwidth adaptation that reduces the expert sizes with tolerable accuracy loss; (2) expert preloading that predicts the activated experts ahead of time and preloads it with the compute-I/O pipeline. On popular MoE LLMs and edge devices, EdgeMoE showcase significant memory savings and speedup over competitive baselines. The code is available at https://github.com/UbiquitousLearning/mllm.
♻ ☆ Dialogue Ontology Relation Extraction via Constrained Chain-of-Thought Decoding SIGDIAL 2024
State-of-the-art task-oriented dialogue systems typically rely on task-specific ontologies for fulfilling user queries. The majority of task-oriented dialogue data, such as customer service recordings, comes without ontology and annotation. Such ontologies are normally built manually, limiting the application of specialised systems. Dialogue ontology construction is an approach for automating that process and typically consists of two steps: term extraction and relation extraction. In this work, we focus on relation extraction in a transfer learning set-up. To improve the generalisation, we propose an extension to the decoding mechanism of large language models. We adapt Chain-of-Thought (CoT) decoding, recently developed for reasoning problems, to generative relation extraction. Here, we generate multiple branches in the decoding space and select the relations based on a confidence threshold. By constraining the decoding to ontology terms and relations, we aim to decrease the risk of hallucination. We conduct extensive experimentation on two widely used datasets and find improvements in performance on target ontology for source fine-tuned and one-shot prompted large language models.
comment: Accepted to appear at SIGDIAL 2024. 9 pages, 4 figures
♻ ☆ Long-Term Auto-Regressive Prediction using Lightweight AI Models: Adams-Bashforth Time Integration with Adaptive Multi-Step Rollout
This study addresses the critical challenge of error accumulation in spatio-temporal auto-regressive predictions within scientific machine learning models by introducing innovative temporal integration schemes and adaptive multi-step rollout strategies. We present a comprehensive analysis of time integration methods, highlighting the adaptation of the two-step Adams-Bashforth scheme to enhance long-term prediction robustness in auto-regressive models. Additionally, we improve temporal prediction accuracy through a multi-step rollout strategy that incorporates multiple future time steps during training, supported by three newly proposed approaches that dynamically adjust the importance of each future step. Despite using an extremely lightweight graph neural network with just 1,177 trainable parameters and training on only 50 snapshots, our framework accurately predicts 350 future time steps (a 7:1 prediction-to-training ratio) achieving an error of only 1.6% compared to the vanilla auto-regressive approach. Moreover, our framework demonstrates an 83% improvement in rollout performance over the standard noise injection method, a standard technique for enhancing long-term rollout performance. Its effectiveness is further validated in more challenging scenarios with truncated meshes, showcasing its adaptability and robustness in practical applications. This work introduces a versatile framework for robust long-term spatio-temporal auto-regressive predictions that shows potential for mitigating error accumulation across various model types and engineering disciplines.
♻ ☆ Safe and Efficient Online Convex Optimization with Linear Budget Constraints and Partial Feedback
This paper studies online convex optimization with unknown linear budget constraints, where only the gradient information of the objective and the bandit feedback of constraint functions are observed. We propose a safe and efficient Lyapunov-optimization algorithm (SELO) that can achieve an $O(\sqrt{T})$ regret and zero cumulative constraint violation. The result also implies SELO achieves $O(\sqrt{T})$ regret when the budget is hard and not allowed to be violated. The proposed algorithm is computationally efficient as it resembles a primal-dual algorithm where the primal problem is an unconstrained, strongly convex and smooth problem, and the dual problem has a simple gradient-type update. The algorithm and theory are further justified in a simulated application of energy-efficient task processing in distributed data centers.
♻ ☆ Fast Robust Kernel Regression through Sign Gradient Descent with Early Stopping
Kernel ridge regression, KRR, is a generalization of linear ridge regression that is non-linear in the data, but linear in the model parameters. Here, we introduce an equivalent formulation of the objective function of KRR, which opens up for replacing the ridge penalty with the $\ell_\infty$ and $\ell_1$ penalties. Using the $\ell_\infty$ and $\ell_1$ penalties, we obtain robust and sparse kernel regression, respectively. We study the similarities between explicitly regularized kernel regression and the solutions obtained by early stopping of iterative gradient-based methods, where we connect $\ell_\infty$ regularization to sign gradient descent, $\ell_1$ regularization to forward stagewise regression (also known as coordinate descent), and $\ell_2$ regularization to gradient descent, and, in the last case, theoretically bound for the differences. We exploit the close relations between $\ell_\infty$ regularization and sign gradient descent, and between $\ell_1$ regularization and coordinate descent to propose computationally efficient methods for robust and sparse kernel regression. We finally compare robust kernel regression through sign gradient descent to existing methods for robust kernel regression on five real data sets, demonstrating that our method is one to two orders of magnitude faster, without compromised accuracy.
comment: Article arXiv:2306.16838v1 has been updated and split into two articles: this article and arXiv:2311.01762. Thus, some of the content in arXiv:2306.16838v1 is not a part of arXiv:2306.16838v2, but of arXiv:2311.01762
♻ ☆ Reward Fine-Tuning Two-Step Diffusion Models via Learning Differentiable Latent-Space Surrogate Reward CVPR 2025
Recent research has shown that fine-tuning diffusion models (DMs) with arbitrary rewards, including non-differentiable ones, is feasible with reinforcement learning (RL) techniques, enabling flexible model alignment. However, applying existing RL methods to timestep-distilled DMs is challenging for ultra-fast ($\le2$-step) image generation. Our analysis suggests several limitations of policy-based RL methods such as PPO or DPO toward this goal. Based on the insights, we propose fine-tuning DMs with learned differentiable surrogate rewards. Our method, named LaSRO, learns surrogate reward models in the latent space of SDXL to convert arbitrary rewards into differentiable ones for efficient reward gradient guidance. LaSRO leverages pre-trained latent DMs for reward modeling and specifically targets image generation $\le2$ steps for reward optimization, enhancing generalizability and efficiency. LaSRO is effective and stable for improving ultra-fast image generation with different reward objectives, outperforming popular RL methods including PPO and DPO. We further show LaSRO's connection to value-based RL, providing theoretical insights. See our webpage at https://sites.google.com/view/lasro.
comment: CVPR 2025
♻ ☆ Learning Force Distribution Estimation for the GelSight Mini Optical Tactile Sensor Based on Finite Element Analysis
Contact-rich manipulation remains a major challenge in robotics. Optical tactile sensors like GelSight Mini offer a low-cost solution for contact sensing by capturing soft-body deformations of the silicone gel. However, accurately inferring shear and normal force distributions from these gel deformations has yet to be fully addressed. In this work, we propose a machine learning approach using a U-net architecture to predict force distributions directly from the sensor's raw images. Our model, trained on force distributions inferred from Finite Element Analysis (FEA), demonstrates promising accuracy in predicting normal and shear force distributions for the commercially available GelSight Mini sensor. It also shows potential for generalization across indenters, sensors of the same type, and for enabling real-time application. The codebase, dataset and models are open-sourced and available at https://feats-ai.github.io .
♻ ☆ Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs
We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-quality web and synthetic data, significantly outperforming recent open-source models of similar size and matching the performance of models twice its size on math and coding tasks requiring complex reasoning. This achievement is driven by a carefully curated synthetic data recipe emphasizing high-quality math and coding datasets. Compared to its predecessor, Phi-3.5-Mini, Phi-4-Mini features an expanded vocabulary size of 200K tokens to better support multilingual applications, as well as group query attention for more efficient long-sequence generation. Phi-4-Multimodal is a multimodal model that integrates text, vision, and speech/audio input modalities into a single model. Its novel modality extension approach leverages LoRA adapters and modality-specific routers to allow multiple inference modes combining various modalities without interference. For example, it now ranks first in the OpenASR leaderboard to date, although the LoRA component of the speech/audio modality has just 460 million parameters. Phi-4-Multimodal supports scenarios involving (vision + language), (vision + speech), and (speech/audio) inputs, outperforming larger vision-language and speech-language models on a wide range of tasks. Additionally, we experiment to further train Phi-4-Mini to enhance its reasoning capabilities. Despite its compact 3.8-billion-parameter size, this experimental version achieves reasoning performance on par with or surpassing significantly larger models, including DeepSeek-R1-Distill-Qwen-7B and DeepSeek-R1-Distill-Llama-8B.
comment: 39 pages
♻ ☆ Transformers for molecular property prediction: Domain adaptation efficiently improves performance
Most of the current transformer-based chemical language models are pre-trained on millions to billions of molecules. However, the improvement from such scaling in dataset size is not confidently linked to improved molecular property prediction. The aim of this study is to investigate and overcome some of the limitations of transformer models in predicting molecular properties. Specifically, we examine the impact of pre-training dataset size and diversity on the performance of transformer models and investigate the use of domain adaptation as a technique for improving model performance. First, our findings indicate that increasing pretraining dataset size beyond 400K molecules from the GuacaMol dataset does not result in a significant improvement on four ADME endpoints, namely, solubility, permeability, microsomal stability, and plasma protein binding. Second, our results demonstrate that using domain adaptation by further training the transformer model on a small set of domain-relevant molecules, i.e., a few hundred to a few thousand, using multi-task regression of physicochemical properties was sufficient to significantly improve performance for three out of the four investigated ADME endpoints (P-value < 0.001). Finally, we observe that a model pre-trained on 400K molecules and domain adopted on a few hundred/thousand molecules performs similarly (P-value > 0.05) to more complicated transformer models like MolBERT(pre-trained on 1.3M molecules) and MolFormer (pre-trained on 100M molecules). A comparison to a random forest model trained on basic physicochemical properties showed similar performance to the examined transformer models. We believe that current transformer models can be improved through further systematic analysis of pre-training and downstream data, pre-training objectives, and scaling laws, ultimately leading to better and more helpful models.
♻ ☆ Generalized moduli of continuity under irregular or random deformations via multiscale analysis
Motivated by the problem of robustness to deformations of the input for deep convolutional neural networks, we identify signal classes which are inherently stable to irregular deformations induced by distortion fields $\tau\in L^\infty(\mathbb{R}^d;\mathbb{R}^d)$, to be characterized in terms of a generalized modulus of continuity associated with the deformation operator. Resorting to ideas of harmonic and multiscale analysis, we prove that for signals in multiresolution approximation spaces $U_s$ at scale $s$, stability in $L^2$ holds in the regime $\|\tau\|_{L^\infty}/s\ll 1$ - essentially as an effect of the uncertainty principle. Instability occurs when $\|\tau\|_{L^\infty}/s\gg 1$, and we provide a sharp upper bound for the asymptotic growth rate. The stability results are then extended to signals in the Besov space $B^{d/2}_{2,1}$ tailored to the given multiresolution approximation. We also consider the case of more general time-frequency deformations. Finally, we provide stochastic versions of the aforementioned results, namely we study the issue of stability in mean when $\tau(x)$ is modeled as a random field (not bounded, in general) with identically distributed variables $|\tau(x)|$, $x\in\mathbb{R}^d$.
comment: 25 pages
♻ ☆ Beyond RMSE and MAE: Introducing EAUC to unmask hidden bias and unfairness in dyadic regression models
Dyadic regression models, which output real-valued predictions for pairs of entities, are fundamental in many domains (e.g. obtaining user-product ratings in Recommender Systems) and promising and under exploration in others (e.g. tuning patient-drug dosages in precision pharmacology). In this work, we prove that non-uniform observed value distributions of individual entities lead to severe biases in state-of-the-art models, skewing predictions towards the average of observed past values for the entity and providing worse-than-random predictive power in eccentric yet crucial cases; we name this phenomenon eccentricity bias. We show that global error metrics like Root Mean Squared Error (RMSE) are insufficient to capture this bias, and we introduce Eccentricity-Area Under the Curve (EAUC) as a novel metric that can quantify it in all studied domains and models. We prove the intuitive interpretation of EAUC by experimenting with naive post-training bias corrections, and theorize other options to use EAUC to guide the construction of fair models. This work contributes a bias-aware evaluation of dyadic regression to prevent unfairness in critical real-world applications of such systems.
♻ ☆ ARIES: Stimulating Self-Refinement of Large Language Models by Iterative Preference Optimization
A truly intelligent Large Language Model (LLM) should be capable of correcting errors in its responses through external interactions. However, even the most advanced models often face challenges in improving their outputs. In this paper, we explore how to cultivate LLMs with the self-refinement capability through iterative preference training, and how this ability can be leveraged to improve model performance during inference. To this end, we introduce a novel post-training and inference framework, called ARIES: Adaptive Refinement and Iterative Enhancement Structure. This method iteratively performs preference training and self-refinement-based data collection. During training, ARIES strengthen the model's direct question-answering capability while simultaneously unlocking its self-refinement potential. During inference, ARIES harnesses this self-refinement capability to generate a series of progressively refined responses, which are then filtered using either the Reward Model Scoring or a simple yet effective Rule-Based Selection mechanism, specifically tailored to our approach, to construct a dataset for the next round of preference training. Experimental results demonstrate the remarkable performance of ARIES. When applied to the Llama-3.1-8B model and under the self-refinement setting, ARIES surpasses powerful models such as GPT-4o, achieving 62.3% length-controlled (LC) and a 63.3% raw win rates on AlpacaEval 2, outperforming Iterative DPO by 27.8% and 35.5% respectively, as well as a 50.3% win rate on Arena-Hard, surpassing Iterative DPO by 26.6%. Furthermore, ARIES consistently enhances performance on mathematical reasoning tasks like GSM8K and MATH.
♻ ☆ CLIP meets DINO for Tuning Zero-Shot Classifier using Unlabeled Image Collections
In the era of foundation models, CLIP has emerged as a powerful tool for aligning text & visual modalities into a common embedding space. However, the alignment objective used to train CLIP often results in subpar visual features for fine-grained tasks. In contrast, SSL-pretrained models like DINO excel at extracting rich visual features due to their specialized training paradigm. Yet, these SSL models require an additional supervised linear probing step, which relies on fully labeled data which is often expensive and difficult to obtain at scale. In this paper, we propose a label-free prompt-tuning method that leverages the rich visual features of self-supervised learning models (DINO) and the broad textual knowledge of large language models (LLMs) to largely enhance CLIP-based image classification performance using unlabeled images. Our approach unfolds in three key steps: (1) We generate robust textual feature embeddings that more accurately represent object classes by leveraging class-specific descriptions from LLMs, enabling more effective zero-shot classification compared to CLIP's default name-specific prompts. (2) These textual embeddings are then used to produce pseudo-labels to train an alignment module that integrates the complementary strengths of LLM description-based textual embeddings & DINO's visual features. (3) Finally, we prompt-tune CLIP's vision encoder through DINO-assisted supervision using the trained alignment module. This three-step process allows us to harness the best of visual & textual foundation models, resulting in a powerful and efficient approach that surpasses state-of-the-art label-free classification methods. Notably, our framework, NoLA (No Labels Attached), achieves an average absolute gain of 3.6% over the state-of-the-art LaFTer across 11 diverse image classification datasets. Our code & models can be found at https://github.com/fazliimam/NoLA.
♻ ☆ Human Implicit Preference-Based Policy Fine-tuning for Multi-Agent Reinforcement Learning in USV Swarm
Multi-Agent Reinforcement Learning (MARL) has shown promise in solving complex problems involving cooperation and competition among agents, such as an Unmanned Surface Vehicle (USV) swarm used in search and rescue, surveillance, and vessel protection. However, aligning system behavior with user preferences is challenging due to the difficulty of encoding expert intuition into reward functions. To address the issue, we propose a Reinforcement Learning with Human Feedback (RLHF) approach for MARL that resolves credit-assignment challenges through an Agent-Level Feedback system categorizing feedback into intra-agent, inter-agent, and intra-team types. To overcome the challenges of direct human feedback, we employ a Large Language Model (LLM) evaluator to validate our approach using feedback scenarios such as region constraints, collision avoidance, and task allocation. Our method effectively refines USV swarm policies, addressing key challenges in multi-agent systems while maintaining fairness and performance consistency.
comment: 7 pages, 4 figures
♻ ☆ A Learned Generalized Geodesic Distance Function-Based Approach for Node Feature Augmentation on Graphs KDD 2024
Geodesic distances on manifolds have numerous applications in image processing, computer graphics and computer vision. In this work, we introduce an approach called `LGGD' (Learned Generalized Geodesic Distances). This method involves generating node features by learning a generalized geodesic distance function through a training pipeline that incorporates training data, graph topology and the node content features. The strength of this method lies in the proven robustness of the generalized geodesic distances to noise and outliers. Our contributions encompass improved performance in node classification tasks, competitive results with state-of-the-art methods on real-world graph datasets, the demonstration of the learnability of parameters within the generalized geodesic equation on graph, and dynamic inclusion of new labels.
comment: Accepted at KDD 2024 Research Track
♻ ☆ Wyckoff Transformer: Generation of Symmetric Crystals
Symmetry rules that atoms obey when they bond together to form an ordered crystal play a fundamental role in determining their physical, chemical, and electronic properties such as electrical and thermal conductivity, optical and polarization behavior, and mechanical strength. Almost all known crystalline materials have internal symmetry. Consistently generating stable crystal structures is still an open challenge, specifically because such symmetry rules are not accounted for. To address this issue, we propose WyFormer, a generative model for materials conditioned on space group symmetry. We use Wyckoff positions as the basis for an elegant, compressed, and discrete structure representation. To model the distribution, we develop a permutation-invariant autoregressive model based on the Transformer and an absence of positional encoding. WyFormer has a unique and powerful synergy of attributes, proven by extensive experimentation: best-in-class symmetry-conditioned generation, physics-motivated inductive bias, competitive stability of the generated structures, competitive material property prediction quality, and unparalleled inference speed.
comment: https://github.com/SymmetryAdvantage/WyckoffTransformer
♻ ☆ BigMac: A Communication-Efficient Mixture-of-Experts Model Structure for Fast Training and Inference
The Mixture-of-Experts (MoE) structure scales the Transformer-based large language models (LLMs) and improves their performance with only the sub-linear increase in computation resources. Recently, a fine-grained DeepSeekMoE structure is proposed, which can further improve the computing efficiency of MoE without performance degradation. However, the All-to-All communication introduced by MoE has become a bottleneck, especially for the fine-grained structure, which typically involves and activates more experts, hence contributing to heavier communication overhead. In this paper, we propose a novel MoE structure named BigMac, which is also fine-grained but with high communication efficiency. The innovation of BigMac is mainly due to that we abandon the \textbf{c}ommunicate-\textbf{d}escend-\textbf{a}scend-\textbf{c}ommunicate (CDAC) manner used by fine-grained MoE, which leads to the All-to-All communication always taking place at the highest dimension. Instead, BigMac designs an efficient \textbf{d}escend-\textbf{c}ommunicate-\textbf{c}ommunicate-\textbf{a}scend (DCCA) manner. Specifically, we add a descending and ascending projection at the entrance and exit of the expert, respectively, which enables the communication to perform at a very low dimension. Furthermore, to adapt to DCCA, we re-design the structure of small experts, ensuring that the expert in BigMac has enough complexity to address tokens. Experimental results show that BigMac achieves comparable or even better model quality than fine-grained MoEs with the same number of experts and a similar number of total parameters. Equally importantly, BigMac reduces the end-to-end latency by up to 3.09$\times$ for training and increases the throughput by up to 3.11$\times$ for inference on state-of-the-art AI computing frameworks including Megatron, Tutel, and DeepSpeed-Inference.
comment: Typo Fixed
♻ ☆ Meta Curvature-Aware Minimization for Domain Generalization
Domain generalization (DG) aims to enhance the ability of models trained on source domains to generalize effectively to unseen domains. Recently, Sharpness-Aware Minimization (SAM) has shown promise in this area by reducing the sharpness of the loss landscape to obtain more generalized models. However, SAM and its variants sometimes fail to guide the model toward a flat minimum, and their training processes exhibit limitations, hindering further improvements in model generalization. In this paper, we first propose an improved model training process aimed at encouraging the model to converge to a flat minima. To achieve this, we design a curvature metric that has a minimal effect when the model is far from convergence but becomes increasingly influential in indicating the curvature of the minima as the model approaches a local minimum. Then we derive a novel algorithm from this metric, called Meta Curvature-Aware Minimization (MeCAM), to minimize the curvature around the local minima. Specifically, the optimization objective of MeCAM simultaneously minimizes the regular training loss, the surrogate gap of SAM, and the surrogate gap of meta-learning. We provide theoretical analysis on MeCAM's generalization error and convergence rate, and demonstrate its superiority over existing DG methods through extensive experiments on five benchmark DG datasets, including PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet. Code will be available on GitHub.
comment: 22 pages, 5 figures, 16 tables
♻ ☆ Chip Placement with Diffusion Models
Macro placement is a vital step in digital circuit design that defines the physical location of large collections of components, known as macros, on a 2D chip. Because key performance metrics of the chip are determined by the placement, optimizing it is crucial. Existing learning-based methods typically fall short because of their reliance on reinforcement learning (RL), which is slow and struggles to generalize, requiring online training on each new circuit. Instead, we train a diffusion model capable of placing new circuits zero-shot, using guided sampling in lieu of RL to optimize placement quality. To enable such models to train at scale, we designed a capable yet efficient architecture for the denoising model, and propose a novel algorithm to generate large synthetic datasets for pre-training. To allow zero-shot transfer to real circuits, we empirically study the design decisions of our dataset generation algorithm, and identify several key factors enabling generalization. When trained on our synthetic data, our models generate high-quality placements on unseen, realistic circuits, achieving competitive performance on placement benchmarks compared to state-of-the-art methods.
♻ ☆ When Large Language Models Meet Evolutionary Algorithms: Potential Enhancements and Challenges
Pre-trained large language models (LLMs) exhibit powerful capabilities for generating natural text. Evolutionary algorithms (EAs) can discover diverse solutions to complex real-world problems. Motivated by the common collective and directionality of text generation and evolution, this paper first illustrates the conceptual parallels between LLMs and EAs at a micro level, which includes multiple one-to-one key characteristics: token representation and individual representation, position encoding and fitness shaping, position embedding and selection, Transformers block and reproduction, and model training and parameter adaptation. These parallels highlight potential opportunities for technical advancements in both LLMs and EAs. Subsequently, we analyze existing interdisciplinary research from a macro perspective to uncover critical challenges, with a particular focus on evolutionary fine-tuning and LLM-enhanced EAs. These analyses not only provide insights into the evolutionary mechanisms behind LLMs but also offer potential directions for enhancing the capabilities of artificial agents.
comment: The article has been accepted for publication in Research
♻ ☆ Personalized Hierarchical Split Federated Learning in Wireless Networks
Extreme resource constraints make large-scale machine learning (ML) with distributed clients challenging in wireless networks. On the one hand, large-scale ML requires massive information exchange between clients and server(s). On the other hand, these clients have limited battery and computation powers that are often dedicated to operational computations. Split federated learning (SFL) is emerging as a potential solution to mitigate these challenges, by splitting the ML model into client-side and server-side model blocks, where only the client-side block is trained on the client device. However, practical applications require personalized models that are suitable for the client's personal task. Motivated by this, we propose a personalized hierarchical split federated learning (PHSFL) algorithm that is specially designed to achieve better personalization performance. More specially, owing to the fact that regardless of the severity of the statistical data distributions across the clients, many of the features have similar attributes, we only train the body part of the federated learning (FL) model while keeping the (randomly initialized) classifier frozen during the training phase. We first perform extensive theoretical analysis to understand the impact of model splitting and hierarchical model aggregations on the global model. Once the global model is trained, we fine-tune each client classifier to obtain the personalized models. Our empirical findings suggest that while the globally trained model with the untrained classifier performs quite similarly to other existing solutions, the fine-tuned models show significantly improved personalized performance.
comment: Accepted for publication in IEEE ICC 2025
♻ ☆ Closed-Loop Open-Vocabulary Mobile Manipulation with GPT-4V ICRA
Autonomous robot navigation and manipulation in open environments require reasoning and replanning with closed-loop feedback. In this work, we present COME-robot, the first closed-loop robotic system utilizing the GPT-4V vision-language foundation model for open-ended reasoning and adaptive planning in real-world scenarios.COME-robot incorporates two key innovative modules: (i) a multi-level open-vocabulary perception and situated reasoning module that enables effective exploration of the 3D environment and target object identification using commonsense knowledge and situated information, and (ii) an iterative closed-loop feedback and restoration mechanism that verifies task feasibility, monitors execution success, and traces failure causes across different modules for robust failure recovery. Through comprehensive experiments involving 8 challenging real-world mobile and tabletop manipulation tasks, COME-robot demonstrates a significant improvement in task success rate (~35%) compared to state-of-the-art methods. We further conduct comprehensive analyses to elucidate how COME-robot's design facilitates failure recovery, free-form instruction following, and long-horizon task planning.
comment: 6 pages, Accepted at 2025 IEEE ICRA, website: https://come-robot.github.io/
♻ ☆ FoundationStereo: Zero-Shot Stereo Matching CVPR 2025
Tremendous progress has been made in deep stereo matching to excel on benchmark datasets through per-domain fine-tuning. However, achieving strong zero-shot generalization - a hallmark of foundation models in other computer vision tasks - remains challenging for stereo matching. We introduce FoundationStereo, a foundation model for stereo depth estimation designed to achieve strong zero-shot generalization. To this end, we first construct a large-scale (1M stereo pairs) synthetic training dataset featuring large diversity and high photorealism, followed by an automatic self-curation pipeline to remove ambiguous samples. We then design a number of network architecture components to enhance scalability, including a side-tuning feature backbone that adapts rich monocular priors from vision foundation models to mitigate the sim-to-real gap, and long-range context reasoning for effective cost volume filtering. Together, these components lead to strong robustness and accuracy across domains, establishing a new standard in zero-shot stereo depth estimation. Project page: https://nvlabs.github.io/FoundationStereo/
comment: CVPR 2025
♻ ☆ Preempting Text Sanitization Utility in Resource-Constrained Privacy-Preserving LLM Interactions
Interactions with online Large Language Models raise privacy issues where providers can gather sensitive information about users and their companies from the prompts. While Differential Privacy can be applied on textual prompts through the Multidimensional Laplace Mechanism, we show that it is difficult to anticipate the utility of such sanitized prompt. Poor utility has clear monetary consequences for LLM services charging on a pay-per-use model as well as great amount of computing resources wasted. To this end, we propose an architecture to predict the utility of a given sanitized prompt before it is sent to the LLM. We experimentally show that our architecture helps prevent such resource waste for up to 12% of the prompts. We also reproduce experiments from one of the most cited paper on distance-based DP for text sanitization and show that a potential performance-driven implementation choice completely changes the output while not being explicitly defined in the paper.
♻ ☆ Improved Online Confidence Bounds for Multinomial Logistic Bandits
In this paper, we propose an improved online confidence bound for multinomial logistic (MNL) models and apply this result to MNL bandits, achieving variance-dependent optimal regret. Recently, Lee & Oh (2024) established an online confidence bound for MNL models and achieved nearly minimax-optimal regret in MNL bandits. However, their results still depend on the norm-boundedness of the unknown parameter $B$ and the maximum size of possible outcomes $K$. To address this, we first derive an online confidence bound of $O\left(\sqrt{d \log t} + B \right)$, which is a significant improvement over the previous bound of $O (B \sqrt{d} \log t \log K )$ (Lee & Oh, 2024). This is mainly achieved by establishing tighter self-concordant properties of the MNL loss and introducing a novel intermediary term to bound the estimation error. Using this new online confidence bound, we propose a constant-time algorithm, OFU-MNL++, which achieves a variance-dependent regret bound of $O \Big( d \log T \sqrt{ \sum_{t=1}^T \sigma_t^2 } \Big) $ for sufficiently large $T$, where $\sigma_t^2$ denotes the variance of the rewards at round $t$, $d$ is the dimension of the contexts, and $T$ is the total number of rounds. Furthermore, we introduce a Maximum Likelihood Estimation (MLE)-based algorithm, OFU-MN$^2$L, which achieves an anytime poly(B)-free regret of $O \Big( d \log (BT) \sqrt{ \sum_{t=1}^T \sigma_t^2 } \Big) $.
comment: Preprint. Under review
♻ ☆ Energy-Based Diffusion Language Models for Text Generation
Despite remarkable progress in autoregressive language models, alternative generative paradigms beyond left-to-right generation are still being actively explored. Discrete diffusion models, with the capacity for parallel generation, have recently emerged as a promising alternative. Unfortunately, these models still underperform the autoregressive counterparts, with the performance gap increasing when reducing the number of sampling steps. Our analysis reveals that this degradation is a consequence of an imperfect approximation used by diffusion models. In this work, we propose Energy-based Diffusion Language Model (EDLM), an energy-based model operating at the full sequence level for each diffusion step, introduced to improve the underlying approximation used by diffusion models. More specifically, we introduce an EBM in a residual form, and show that its parameters can be obtained by leveraging a pretrained autoregressive model or by finetuning a bidirectional transformer via noise contrastive estimation. We also propose an efficient generation algorithm via parallel important sampling. Comprehensive experiments on language modeling benchmarks show that our model can consistently outperform state-of-the-art diffusion models by a significant margin, and approaches autoregressive models' perplexity. We further show that, without any generation performance drop, our framework offers a 1.3$\times$ sampling speedup over existing diffusion models. Reproduced code is available at https://github.com/MinkaiXu/Energy-Diffusion-LLM.
♻ ☆ Nearly Minimax Optimal Regret for Multinomial Logistic Bandit NeurIPS 2024
In this paper, we study the contextual multinomial logit (MNL) bandit problem in which a learning agent sequentially selects an assortment based on contextual information, and user feedback follows an MNL choice model. There has been a significant discrepancy between lower and upper regret bounds, particularly regarding the maximum assortment size $K$. Additionally, the variation in reward structures between these bounds complicates the quest for optimality. Under uniform rewards, where all items have the same expected reward, we establish a regret lower bound of $\Omega(d\sqrt{T/K})$ and propose a constant-time algorithm, OFU-MNL+, that achieves a matching upper bound of $\tilde{O}(d\sqrt{T/K})$. We also provide instance-dependent minimax regret bounds under uniform rewards. Under non-uniform rewards, we prove a lower bound of $\Omega(d\sqrt{T})$ and an upper bound of $\tilde{O}(d\sqrt{T})$, also achievable by OFU-MNL+. Our empirical studies support these theoretical findings. To the best of our knowledge, this is the first work in the contextual MNL bandit literature to prove minimax optimality -- for either uniform or non-uniform reward setting -- and to propose a computationally efficient algorithm that achieves this optimality up to logarithmic factors.
comment: Accepted in NeurIPS 2024
Information Retrieval 8
☆ A Survey of Large Language Model Empowered Agents for Recommendation and Search: Towards Next-Generation Information Retrieval
Information technology has profoundly altered the way humans interact with information. The vast amount of content created, shared, and disseminated online has made it increasingly difficult to access relevant information. Over the past two decades, search and recommendation systems (collectively referred to as information retrieval systems) have evolved significantly to address these challenges. Recent advances in large language models (LLMs) have demonstrated capabilities that surpass human performance in various language-related tasks and exhibit general understanding, reasoning, and decision-making abilities. This paper explores the transformative potential of large language model agents in enhancing search and recommendation systems. We discuss the motivations and roles of LLM agents, and establish a classification framework to elaborate on the existing research. We highlight the immense potential of LLM agents in addressing current challenges in search and recommendation, providing insights into future research directions. This paper is the first to systematically review and classify the research on LLM agents in these domains, offering a novel perspective on leveraging this advanced AI technology for information retrieval. To help understand the existing works, we list the existing papers on agent-based simulation with large language models at this link: https://github.com/tsinghua-fib-lab/LLM-Agent-for-Recommendation-and-Search.
☆ Bridging Classical and Quantum String Matching: A Computational Reformulation of Bit-Parallelism
String matching is a fundamental problem in computer science, with critical applications in text retrieval, bioinformatics, and data analysis. Among the numerous solutions that have emerged for this problem in recent decades, bit-parallelism has significantly enhanced their practical efficiency, leading to the development of several optimized approaches for both exact and approximate string matching. However, their potential in quantum computing remains largely unexplored. This paper presents a novel pathway that not only translates bit-parallel string matching algorithms into the quantum framework but also enhances their performance to achieve a quadratic speedup through Grover's search. By embedding quantum search within a bit-parallel model, we reduce the time complexity of string matching, establishing a structured pathway for transforming classical algorithms into quantum solutions with provable computational advantages. Beyond exact matching, this technique offers a foundation for tackling a wide range of non-standard string matching problems, opening new avenues for efficient text searching in the quantum era. To demonstrate the simplicity and adaptability of the technique presented in this paper, we apply this translation and adaptation process to two landmark bit-parallel algorithms: Shift-And for exact pattern matching and Shift-Add for approximate string matching with up to k errors.
☆ R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning
Existing Large Reasoning Models (LRMs) have shown the potential of reinforcement learning (RL) to enhance the complex reasoning capabilities of Large Language Models~(LLMs). While they achieve remarkable performance on challenging tasks such as mathematics and coding, they often rely on their internal knowledge to solve problems, which can be inadequate for time-sensitive or knowledge-intensive questions, leading to inaccuracies and hallucinations. To address this, we propose \textbf{R1-Searcher}, a novel two-stage outcome-based RL approach designed to enhance the search capabilities of LLMs. This method allows LLMs to autonomously invoke external search systems to access additional knowledge during the reasoning process. Our framework relies exclusively on RL, without requiring process rewards or distillation for a cold start. % effectively generalizing to out-of-domain datasets and supporting both Base and Instruct models. Our experiments demonstrate that our method significantly outperforms previous strong RAG methods, even when compared to the closed-source GPT-4o-mini.
☆ Benchmarking LLMs in Recommendation Tasks: A Comparative Evaluation with Conventional Recommenders
In recent years, integrating large language models (LLMs) into recommender systems has created new opportunities for improving recommendation quality. However, a comprehensive benchmark is needed to thoroughly evaluate and compare the recommendation capabilities of LLMs with traditional recommender systems. In this paper, we introduce RecBench, which systematically investigates various item representation forms (including unique identifier, text, semantic embedding, and semantic identifier) and evaluates two primary recommendation tasks, i.e., click-through rate prediction (CTR) and sequential recommendation (SeqRec). Our extensive experiments cover up to 17 large models and are conducted across five diverse datasets from fashion, news, video, books, and music domains. Our findings indicate that LLM-based recommenders outperform conventional recommenders, achieving up to a 5% AUC improvement in the CTR scenario and up to a 170% NDCG@10 improvement in the SeqRec scenario. However, these substantial performance gains come at the expense of significantly reduced inference efficiency, rendering the LLM-as-RS paradigm impractical for real-time recommendation environments. We aim for our findings to inspire future research, including recommendation-specific model acceleration methods. We will release our code, data, configurations, and platform to enable other researchers to reproduce and build upon our experimental results.
☆ LoRACode: LoRA Adapters for Code Embeddings ICLR 2025
Code embeddings are essential for semantic code search; however, current approaches often struggle to capture the precise syntactic and contextual nuances inherent in code. Open-source models such as CodeBERT and UniXcoder exhibit limitations in scalability and efficiency, while high-performing proprietary systems impose substantial computational costs. We introduce a parameter-efficient fine-tuning method based on Low-Rank Adaptation (LoRA) to construct task-specific adapters for code retrieval. Our approach reduces the number of trainable parameters to less than two percent of the base model, enabling rapid fine-tuning on extensive code corpora (2 million samples in 25 minutes on two H100 GPUs). Experiments demonstrate an increase of up to 9.1% in Mean Reciprocal Rank (MRR) for Code2Code search, and up to 86.69% for Text2Code search tasks across multiple programming languages. Distinction in task-wise and language-wise adaptation helps explore the sensitivity of code retrieval for syntactical and linguistic variations.
comment: Accepted at the Deep Learning for Code (DL4C) Workshop at ICLR 2025
☆ MOHPER: Multi-objective Hyperparameter Optimization Framework for E-commerce Retrieval System
E-commerce search optimization has evolved to include a wider range of metrics that reflect user engagement and business objectives. Modern search frameworks now incorporate advanced quality features, such as sales counts and document-query relevance, to better align search results with these goals. Traditional methods typically focus on click-through rate (CTR) as a measure of engagement or relevance, but this can miss true purchase intent, creating a gap between user interest and actual conversions. Joint training with the click-through conversion rate (CTCVR) has become essential for understanding buying behavior, although its sparsity poses challenges for reliable optimization. This study presents MOHPER, a Multi-Objective Hyperparameter Optimization framework for E-commerce Retrieval systems. Utilizing Bayesian optimization and sampling, it jointly optimizes both CTR, CTCVR, and relevant objectives, focusing on engagement and conversion of the users. In addition, to improve the selection of the best configuration from multi-objective optimization, we suggest advanced methods for hyperparameter selection, including a meta-configuration voting strategy and a cumulative training approach that leverages prior optimal configurations, to improve speeds of training and efficiency. Currently deployed in a live setting, our proposed framework substantiates its practical efficacy in achieving a balanced optimization that aligns with both user satisfaction and revenue goals.
♻ ☆ Sustainable transparency in Recommender Systems: Bayesian Ranking of Images for Explainability
Recommender Systems have become crucial in the modern world, commonly guiding users towards relevant content or products, and having a large influence over the decisions of users and citizens. However, ensuring transparency and user trust in these systems remains a challenge; personalized explanations have emerged as a solution, offering justifications for recommendations. Among the existing approaches for generating personalized explanations, using existing visual content created by users is a promising option to maximize transparency and user trust. State-of-the-art models that follow this approach, despite leveraging highly optimized architectures, employ surrogate learning tasks that do not efficiently model the objective of ranking images as explanations for a given recommendation; this leads to a suboptimal training process with high computational costs that may not be reduced without affecting model performance. This work presents BRIE, a novel model where we leverage Bayesian Pairwise Ranking to enhance the training process, allowing us to consistently outperform state-of-the-art models in six real-world datasets while reducing its model size by up to 64 times and its CO2 emissions by up to 75% in training and inference.
♻ ☆ Beyond RMSE and MAE: Introducing EAUC to unmask hidden bias and unfairness in dyadic regression models
Dyadic regression models, which output real-valued predictions for pairs of entities, are fundamental in many domains (e.g. obtaining user-product ratings in Recommender Systems) and promising and under exploration in others (e.g. tuning patient-drug dosages in precision pharmacology). In this work, we prove that non-uniform observed value distributions of individual entities lead to severe biases in state-of-the-art models, skewing predictions towards the average of observed past values for the entity and providing worse-than-random predictive power in eccentric yet crucial cases; we name this phenomenon eccentricity bias. We show that global error metrics like Root Mean Squared Error (RMSE) are insufficient to capture this bias, and we introduce Eccentricity-Area Under the Curve (EAUC) as a novel metric that can quantify it in all studied domains and models. We prove the intuitive interpretation of EAUC by experimenting with naive post-training bias corrections, and theorize other options to use EAUC to guide the construction of fair models. This work contributes a bias-aware evaluation of dyadic regression to prevent unfairness in critical real-world applications of such systems.
Computation and Language 159
☆ L$^2$M: Mutual Information Scaling Law for Long-Context Language Modeling
We rigorously establish a bipartite mutual information scaling law in natural language that governs long-range dependencies. This scaling law, which we show is distinct from and scales independently of the conventional two-point mutual information, is the key to understanding long-context language modeling. Using this scaling law, we formulate the Long-context Language Modeling (L$^2$M) condition, which relates a model's capacity for effective long context length modeling to the scaling of its latent state size for storing past information. Our results are validated through experiments on both transformers and state space models. This work establishes a theoretical foundation that guides the development of large language models toward longer context lengths.
comment: 29 pages, 12 figures, 1 table
LLMVoX: Autoregressive Streaming Text-to-Speech Model for Any LLM
Recent advancements in speech-to-speech dialogue systems leverage LLMs for multimodal interactions, yet they remain hindered by fine-tuning requirements, high computational overhead, and text-speech misalignment. Existing speech-enabled LLMs often degrade conversational quality by modifying the LLM, thereby compromising its linguistic capabilities. In contrast, we propose LLMVoX, a lightweight 30M-parameter, LLM-agnostic, autoregressive streaming TTS system that generates high-quality speech with low latency, while fully preserving the capabilities of the base LLM. Our approach achieves a significantly lower Word Error Rate compared to speech-enabled LLMs, while operating at comparable latency and UTMOS score. By decoupling speech synthesis from LLM processing via a multi-queue token streaming system, LLMVoX supports seamless, infinite-length dialogues. Its plug-and-play design also facilitates extension to various tasks with different backbones. Furthermore, LLMVoX generalizes to new languages with only dataset adaptation, attaining a low Character Error Rate on an Arabic speech task. Additionally, we have integrated LLMVoX with a Vision-Language Model to create an omni-model with speech, text, and vision capabilities, without requiring additional multimodal training. Our code base and project page is available at https://mbzuai-oryx.github.io/LLMVoX .
☆ Shifting Long-Context LLMs Research from Input to Output
Recent advancements in long-context Large Language Models (LLMs) have primarily concentrated on processing extended input contexts, resulting in significant strides in long-context comprehension. However, the equally critical aspect of generating long-form outputs has received comparatively less attention. This paper advocates for a paradigm shift in NLP research toward addressing the challenges of long-output generation. Tasks such as novel writing, long-term planning, and complex reasoning require models to understand extensive contexts and produce coherent, contextually rich, and logically consistent extended text. These demands highlight a critical gap in current LLM capabilities. We underscore the importance of this under-explored domain and call for focused efforts to develop foundational LLMs tailored for generating high-quality, long-form outputs, which hold immense potential for real-world applications.
comment: Preprint
☆ Enough Coin Flips Can Make LLMs Act Bayesian
Large language models (LLMs) exhibit the ability to generalize given few-shot examples in their input prompt, an emergent capability known as in-context learning (ICL). We investigate whether LLMs utilize ICL to perform structured reasoning in ways that are consistent with a Bayesian framework or rely on pattern matching. Using a controlled setting of biased coin flips, we find that: (1) LLMs often possess biased priors, causing initial divergence in zero-shot settings, (2) in-context evidence outweighs explicit bias instructions, (3) LLMs broadly follow Bayesian posterior updates, with deviations primarily due to miscalibrated priors rather than flawed updates, and (4) attention magnitude has negligible effect on Bayesian inference. With sufficient demonstrations of biased coin flips via ICL, LLMs update their priors in a Bayesian manner.
☆ Full-Duplex-Bench: A Benchmark to Evaluate Full-duplex Spoken Dialogue Models on Turn-taking Capabilities
Spoken dialogue modeling introduces unique challenges beyond text-based language modeling, demanding robust turn-taking, backchanneling, and real-time interaction. Although most Spoken Dialogue Models (SDMs) rely on half-duplex processing (handling speech one turn at a time), emerging full-duplex SDMs can listen and speak simultaneously, enabling more natural and engaging conversations. However, current evaluations of such models remain limited, often focusing on turn-based metrics or high-level corpus analyses (e.g., turn gaps, pauses). To address this gap, we present Full-Duplex-Bench, a new benchmark that systematically evaluates key conversational behaviors: pause handling, backchanneling, turn-taking, and interruption management. Our framework uses automatic metrics for consistent and reproducible assessments of SDMs' interactive performance. By offering an open and standardized evaluation benchmark, we aim to advance spoken dialogue modeling and encourage the development of more interactive and natural dialogue systems.
☆ Scaling Rich Style-Prompted Text-to-Speech Datasets
We introduce Paralinguistic Speech Captions (ParaSpeechCaps), a large-scale dataset that annotates speech utterances with rich style captions. While rich abstract tags (e.g. guttural, nasal, pained) have been explored in small-scale human-annotated datasets, existing large-scale datasets only cover basic tags (e.g. low-pitched, slow, loud). We combine off-the-shelf text and speech embedders, classifiers and an audio language model to automatically scale rich tag annotations for the first time. ParaSpeechCaps covers a total of 59 style tags, including both speaker-level intrinsic tags and utterance-level situational tags. It consists of 342 hours of human-labelled data (PSC-Base) and 2427 hours of automatically annotated data (PSC-Scaled). We finetune Parler-TTS, an open-source style-prompted TTS model, on ParaSpeechCaps, and achieve improved style consistency (+7.9% Consistency MOS) and speech quality (+15.5% Naturalness MOS) over the best performing baseline that combines existing rich style tag datasets. We ablate several of our dataset design choices to lay the foundation for future work in this space. Our dataset, models and code are released at https://github.com/ajd12342/paraspeechcaps .
☆ L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning
Reasoning language models have shown an uncanny ability to improve performance at test-time by ``thinking longer''-that is, by generating longer chain-of-thought sequences and hence using more compute. However, the length of their chain-of-thought reasoning is not controllable, making it impossible to allocate test-time compute to achieve a desired level of performance. We introduce Length Controlled Policy Optimization (LCPO), a simple reinforcement learning method that optimizes for accuracy and adherence to user-specified length constraints. We use LCPO to train L1, a reasoning language model that produces outputs satisfying a length constraint given in its prompt. L1's length control allows for smoothly trading off computational cost and accuracy on a wide range of tasks, and outperforms the state-of-the-art S1 method for length control. Furthermore, we uncover an unexpected short chain-of-thought capability in models trained with LCPO. For instance, our 1.5B L1 model surpasses GPT-4o at equal reasoning lengths. Overall, LCPO enables precise control over reasoning length, allowing for fine-grained allocation of test-time compute and accuracy. We release code and models at https://www.cmu-l3.github.io/l1
☆ UIPE: Enhancing LLM Unlearning by Removing Knowledge Related to Forgetting Targets
Large Language Models (LLMs) inevitably acquire harmful information during training on massive datasets. LLM unlearning aims to eliminate the influence of such harmful information while maintaining the model's overall performance. Existing unlearning methods, represented by gradient ascent-based approaches, primarily focus on forgetting target data while overlooking the crucial impact of logically related knowledge on the effectiveness of unlearning. In this paper, through both theoretical and experimental analyses, we first demonstrate that a key reason for the suboptimal unlearning performance is that models can reconstruct the target content through reasoning with logically related knowledge. To address this issue, we propose Unlearning Improvement via Parameter Extrapolation (UIPE), a method that removes knowledge highly correlated with the forgetting targets. Experimental results show that UIPE significantly enhances the performance of various mainstream LLM unlearning methods on the TOFU benchmark.
☆ Quantifying the Reasoning Abilities of LLMs on Real-world Clinical Cases
The latest reasoning-enhanced large language models (reasoning LLMs), such as DeepSeek-R1 and OpenAI-o3, have demonstrated remarkable success. However, the application of such reasoning enhancements to the highly professional medical domain has not been clearly evaluated, particularly regarding with not only assessing the final generation but also examining the quality of their reasoning processes. In this study, we present MedR-Bench, a reasoning-focused medical evaluation benchmark comprising 1,453 structured patient cases with reasoning references mined from case reports. Our benchmark spans 13 body systems and 10 specialty disorders, encompassing both common and rare diseases. In our evaluation, we introduce a versatile framework consisting of three critical clinical stages: assessment recommendation, diagnostic decision-making, and treatment planning, comprehensively capturing the LLMs' performance across the entire patient journey in healthcare. For metrics, we propose a novel agentic system, Reasoning Evaluator, designed to automate and objectively quantify free-text reasoning responses in a scalable manner from the perspectives of efficiency, factuality, and completeness by dynamically searching and performing cross-referencing checks. As a result, we assess five state-of-the-art reasoning LLMs, including DeepSeek-R1, OpenAI-o3-mini, and others. Our results reveal that current LLMs can handle relatively simple diagnostic tasks with sufficient critical assessment results, achieving accuracy generally over 85%. However, they still struggle with more complex tasks, such as assessment recommendation and treatment planning. In reasoning, their reasoning processes are generally reliable, with factuality scores exceeding 90%, though they often omit critical reasoning steps. Our study clearly reveals further development directions for current clinical LLMs.
☆ DIMSUM: Discourse in Mathematical Reasoning as a Supervision Module
We look at reasoning on GSM8k, a dataset of short texts presenting primary school, math problems. We find, with Mirzadeh et al. (2024), that current LLM progress on the data set may not be explained by better reasoning but by exposure to a broader pretraining data distribution. We then introduce a novel information source for helping models with less data or inferior training reason better: discourse structure. We show that discourse structure improves performance for models like Llama2 13b by up to 160%. Even for models that have most likely memorized the data set, adding discourse structural information to the model still improves predictions and dramatically improves large model performance on out of distribution examples.
LLM-guided Plan and Retrieval: A Strategic Alignment for Interpretable User Satisfaction Estimation in Dialogue NAACL 2025
Understanding user satisfaction with conversational systems, known as User Satisfaction Estimation (USE), is essential for assessing dialogue quality and enhancing user experiences. However, existing methods for USE face challenges due to limited understanding of underlying reasons for user dissatisfaction and the high costs of annotating user intentions. To address these challenges, we propose PRAISE (Plan and Retrieval Alignment for Interpretable Satisfaction Estimation), an interpretable framework for effective user satisfaction prediction. PRAISE operates through three key modules. The Strategy Planner develops strategies, which are natural language criteria for classifying user satisfaction. The Feature Retriever then incorporates knowledge on user satisfaction from Large Language Models (LLMs) and retrieves relevance features from utterances. Finally, the Score Analyzer evaluates strategy predictions and classifies user satisfaction. Experimental results demonstrate that PRAISE achieves state-of-the-art performance on three benchmarks for the USE task. Beyond its superior performance, PRAISE offers additional benefits. It enhances interpretability by providing instance-level explanations through effective alignment of utterances with strategies. Moreover, PRAISE operates more efficiently than existing approaches by eliminating the need for LLMs during the inference phase.
comment: Accepted by NAACL 2025
☆ An Information-theoretic Multi-task Representation Learning Framework for Natural Language Understanding AAAI 2025
This paper proposes a new principled multi-task representation learning framework (InfoMTL) to extract noise-invariant sufficient representations for all tasks. It ensures sufficiency of shared representations for all tasks and mitigates the negative effect of redundant features, which can enhance language understanding of pre-trained language models (PLMs) under the multi-task paradigm. Firstly, a shared information maximization principle is proposed to learn more sufficient shared representations for all target tasks. It can avoid the insufficiency issue arising from representation compression in the multi-task paradigm. Secondly, a task-specific information minimization principle is designed to mitigate the negative effect of potential redundant features in the input for each task. It can compress task-irrelevant redundant information and preserve necessary information relevant to the target for multi-task prediction. Experiments on six classification benchmarks show that our method outperforms 12 comparative multi-task methods under the same multi-task settings, especially in data-constrained and noisy scenarios. Extensive experiments demonstrate that the learned representations are more sufficient, data-efficient, and robust.
comment: 11 pages, accepted to AAAI 2025 (main conference), the code is available at https://github.com/zerohd4869/InfoMTL
☆ Implicit Cross-Lingual Rewarding for Efficient Multilingual Preference Alignment
Direct Preference Optimization (DPO) has become a prominent method for aligning Large Language Models (LLMs) with human preferences. While DPO has enabled significant progress in aligning English LLMs, multilingual preference alignment is hampered by data scarcity. To address this, we propose a novel approach that $\textit{captures}$ learned preferences from well-aligned English models by implicit rewards and $\textit{transfers}$ them to other languages through iterative training. Specifically, we derive an implicit reward model from the logits of an English DPO-aligned model and its corresponding reference model. This reward model is then leveraged to annotate preference relations in cross-lingual instruction-following pairs, using English instructions to evaluate multilingual responses. The annotated data is subsequently used for multilingual DPO fine-tuning, facilitating preference knowledge transfer from English to other languages. Fine-tuning Llama3 for two iterations resulted in a 12.72% average improvement in Win Rate and a 5.97% increase in Length Control Win Rate across all training languages on the X-AlpacaEval leaderboard. Our findings demonstrate that leveraging existing English-aligned models can enable efficient and effective multilingual preference alignment, significantly reducing the need for extensive multilingual preference data. The code is available at https://github.com/ZNLP/Implicit-Cross-Lingual-Rewarding
comment: Work in progress
☆ IFIR: A Comprehensive Benchmark for Evaluating Instruction-Following in Expert-Domain Information Retrieval NAACL 2025
We introduce IFIR, the first comprehensive benchmark designed to evaluate instruction-following information retrieval (IR) in expert domains. IFIR includes 2,426 high-quality examples and covers eight subsets across four specialized domains: finance, law, healthcare, and science literature. Each subset addresses one or more domain-specific retrieval tasks, replicating real-world scenarios where customized instructions are critical. IFIR enables a detailed analysis of instruction-following retrieval capabilities by incorporating instructions at different levels of complexity. We also propose a novel LLM-based evaluation method to provide a more precise and reliable assessment of model performance in following instructions. Through extensive experiments on 15 frontier retrieval models, including those based on LLMs, our results reveal that current models face significant challenges in effectively following complex, domain-specific instructions. We further provide in-depth analyses to highlight these limitations, offering valuable insights to guide future advancements in retriever development.
comment: NAACL 2025 Main
☆ Mark Your LLM: Detecting the Misuse of Open-Source Large Language Models via Watermarking ICLR 2025
As open-source large language models (LLMs) like Llama3 become more capable, it is crucial to develop watermarking techniques to detect their potential misuse. Existing watermarking methods either add watermarks during LLM inference, which is unsuitable for open-source LLMs, or primarily target classification LLMs rather than recent generative LLMs. Adapting these watermarks to open-source LLMs for misuse detection remains an open challenge. This work defines two misuse scenarios for open-source LLMs: intellectual property (IP) violation and LLM Usage Violation. Then, we explore the application of inference-time watermark distillation and backdoor watermarking in these contexts. We propose comprehensive evaluation methods to assess the impact of various real-world further fine-tuning scenarios on watermarks and the effect of these watermarks on LLM performance. Our experiments reveal that backdoor watermarking could effectively detect IP Violation, while inference-time watermark distillation is applicable in both scenarios but less robust to further fine-tuning and has a more significant impact on LLM performance compared to backdoor watermarking. Exploring more advanced watermarking methods for open-source LLMs to detect their misuse should be an important future direction.
comment: Accepted by the 1st Workshop on GenAI Watermarking, collocated with ICLR 2025
SurveyForge: On the Outline Heuristics, Memory-Driven Generation, and Multi-dimensional Evaluation for Automated Survey Writing
Survey paper plays a crucial role in scientific research, especially given the rapid growth of research publications. Recently, researchers have begun using LLMs to automate survey generation for better efficiency. However, the quality gap between LLM-generated surveys and those written by human remains significant, particularly in terms of outline quality and citation accuracy. To close these gaps, we introduce SurveyForge, which first generates the outline by analyzing the logical structure of human-written outlines and referring to the retrieved domain-related articles. Subsequently, leveraging high-quality papers retrieved from memory by our scholar navigation agent, SurveyForge can automatically generate and refine the content of the generated article. Moreover, to achieve a comprehensive evaluation, we construct SurveyBench, which includes 100 human-written survey papers for win-rate comparison and assesses AI-generated survey papers across three dimensions: reference, outline, and content quality. Experiments demonstrate that SurveyForge can outperform previous works such as AutoSurvey.
comment: Code and dataset are available for downloading at: https://github.com/Alpha-Innovator/SurveyForge 22 pages, 10 figures
☆ START: Self-taught Reasoner with Tools
Large reasoning models (LRMs) like OpenAI-o1 and DeepSeek-R1 have demonstrated remarkable capabilities in complex reasoning tasks through the utilization of long Chain-of-thought (CoT). However, these models often suffer from hallucinations and inefficiencies due to their reliance solely on internal reasoning processes. In this paper, we introduce START (Self-Taught Reasoner with Tools), a novel tool-integrated long CoT reasoning LLM that significantly enhances reasoning capabilities by leveraging external tools. Through code execution, START is capable of performing complex computations, self-checking, exploring diverse methods, and self-debugging, thereby addressing the limitations of LRMs. The core innovation of START lies in its self-learning framework, which comprises two key techniques: 1) Hint-infer: We demonstrate that inserting artificially designed hints (e.g., ``Wait, maybe using Python here is a good idea.'') during the inference process of a LRM effectively stimulates its ability to utilize external tools without the need for any demonstration data. Hint-infer can also serve as a simple and effective sequential test-time scaling method; 2) Hint Rejection Sampling Fine-Tuning (Hint-RFT): Hint-RFT combines Hint-infer and RFT by scoring, filtering, and modifying the reasoning trajectories with tool invocation generated by a LRM via Hint-infer, followed by fine-tuning the LRM. Through this framework, we have fine-tuned the QwQ-32B model to achieve START. On PhD-level science QA (GPQA), competition-level math benchmarks (AMC23, AIME24, AIME25), and the competition-level code benchmark (LiveCodeBench), START achieves accuracy rates of 63.6%, 95.0%, 66.7%, 47.1%, and 47.3%, respectively. It significantly outperforms the base QwQ-32B and achieves performance comparable to the state-of-the-art open-weight model R1-Distill-Qwen-32B and the proprietary model o1-Preview.
comment: 38 pages, 5 figures and 6 tables
☆ SynGraph: A Dynamic Graph-LLM Synthesis Framework for Sparse Streaming User Sentiment Modeling
User reviews on e-commerce platforms exhibit dynamic sentiment patterns driven by temporal and contextual factors. Traditional sentiment analysis methods focus on static reviews, failing to capture the evolving temporal relationship between user sentiment rating and textual content. Sentiment analysis on streaming reviews addresses this limitation by modeling and predicting the temporal evolution of user sentiments. However, it suffers from data sparsity, manifesting in temporal, spatial, and combined forms. In this paper, we introduce SynGraph, a novel framework designed to address data sparsity in sentiment analysis on streaming reviews. SynGraph alleviates data sparsity by categorizing users into mid-tail, long-tail, and extreme scenarios and incorporating LLM-augmented enhancements within a dynamic graph-based structure. Experiments on real-world datasets demonstrate its effectiveness in addressing sparsity and improving sentiment modeling in streaming reviews.
comment: 18 pages, 17 figures
☆ Better Process Supervision with Bi-directional Rewarding Signals
Process supervision, i.e., evaluating each step, is critical for complex large language model (LLM) reasoning and test-time searching with increased inference compute. Existing approaches, represented by process reward models (PRMs), primarily focus on rewarding signals up to the current step, exhibiting a one-directional nature and lacking a mechanism to model the distance to the final target. To address this problem, we draw inspiration from the A* algorithm, which states that an effective supervisory signal should simultaneously consider the incurred cost and the estimated cost for reaching the target. Building on this key insight, we introduce BiRM, a novel process supervision model that not only evaluates the correctness of previous steps but also models the probability of future success. We conduct extensive experiments on mathematical reasoning tasks and demonstrate that BiRM provides more precise evaluations of LLM reasoning steps, achieving an improvement of 3.1% on Gaokao2023 over PRM under the Best-of-N sampling method. Besides, in search-based strategies, BiRM provides more comprehensive guidance and outperforms ORM by 5.0% and PRM by 3.8% respectively on MATH-500.
☆ HalluCounter: Reference-free LLM Hallucination Detection in the Wild!
Response consistency-based, reference-free hallucination detection (RFHD) methods do not depend on internal model states, such as generation probabilities or gradients, which Grey-box models typically rely on but are inaccessible in closed-source LLMs. However, their inability to capture query-response alignment patterns often results in lower detection accuracy. Additionally, the lack of large-scale benchmark datasets spanning diverse domains remains a challenge, as most existing datasets are limited in size and scope. To this end, we propose HalluCounter, a novel reference-free hallucination detection method that utilizes both response-response and query-response consistency and alignment patterns. This enables the training of a classifier that detects hallucinations and provides a confidence score and an optimal response for user queries. Furthermore, we introduce HalluCounterEval, a benchmark dataset comprising both synthetically generated and human-curated samples across multiple domains. Our method outperforms state-of-the-art approaches by a significant margin, achieving over 90\% average confidence in hallucination detection across datasets.
comment: 30 pages, 4 figures
☆ Towards Data-Efficient Language Models: A Child-Inspired Approach to Language Learning
In this work, we explain our approach employed in the BabyLM Challenge, which uses various methods of training language models (LMs) with significantly less data compared to traditional large language models (LLMs) and are inspired by how human children learn. While a human child is exposed to far less linguistic input than an LLM, they still achieve remarkable language understanding and generation abilities. To this end, we develop a model trained on a curated dataset consisting of 10 million words, primarily sourced from child-directed transcripts. The 2024 BabyLM Challenge initial dataset of 10M words is filtered to 8.5M. Next, it is supplemented with a randomly selected subset of TVR dataset consisting of 1.5M words of television dialogues. The latter dataset ensures that similar to children, the model is also exposed to language through media. Furthermore, we reduce the vocabulary size to 32,000 tokens, aligning it with the limited vocabulary of children in the early stages of language acquisition. We use curriculum learning and is able to match the baseline on certain benchmarks while surpassing the baseline on others. Additionally, incorporating common LLM training datasets, such as MADLAD-400, degrades performance. These findings underscore the importance of dataset selection, vocabulary scaling, and curriculum learning in creating more data-efficient language models that better mimic human learning processes.
comment: 5 pages
☆ The Best of Both Worlds: Integrating Language Models and Diffusion Models for Video Generation
Recent advancements in text-to-video (T2V) generation have been driven by two competing paradigms: autoregressive language models and diffusion models. However, each paradigm has intrinsic limitations: language models struggle with visual quality and error accumulation, while diffusion models lack semantic understanding and causal modeling. In this work, we propose LanDiff, a hybrid framework that synergizes the strengths of both paradigms through coarse-to-fine generation. Our architecture introduces three key innovations: (1) a semantic tokenizer that compresses 3D visual features into compact 1D discrete representations through efficient semantic compression, achieving a $\sim$14,000$\times$ compression ratio; (2) a language model that generates semantic tokens with high-level semantic relationships; (3) a streaming diffusion model that refines coarse semantics into high-fidelity videos. Experiments show that LanDiff, a 5B model, achieves a score of 85.43 on the VBench T2V benchmark, surpassing the state-of-the-art open-source models Hunyuan Video (13B) and other commercial models such as Sora, Keling, and Hailuo. Furthermore, our model also achieves state-of-the-art performance in long video generation, surpassing other open-source models in this field. Our demo can be viewed at https://landiff.github.io/.
☆ HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Transformers have become the de facto architecture for a wide range of machine learning tasks, particularly in large language models (LLMs). Despite their remarkable performance, challenges remain in training deep transformer networks, especially regarding the location of layer normalization. While Pre-Norm structures facilitate easier training due to their more prominent identity path, they often yield suboptimal performance compared to Post-Norm. In this paper, we propose $\textbf{HybridNorm}$, a straightforward yet effective hybrid normalization strategy that integrates the advantages of both Pre-Norm and Post-Norm approaches. Specifically, HybridNorm employs QKV normalization within the attention mechanism and Post-Norm in the feed-forward network (FFN) of each transformer block. This design not only stabilizes training but also enhances performance, particularly in the context of LLMs. Comprehensive experiments in both dense and sparse architectures show that HybridNorm consistently outperforms both Pre-Norm and Post-Norm approaches, achieving state-of-the-art results across various benchmarks. These findings highlight the potential of HybridNorm as a more stable and effective technique for improving the training and performance of deep transformer models. %Code will be made publicly available. Code is available at https://github.com/BryceZhuo/HybridNorm.
☆ Compositional Causal Reasoning Evaluation in Language Models
Causal reasoning and compositional reasoning are two core aspirations in generative AI. Measuring the extent of these behaviors requires principled evaluation methods. We explore a unified perspective that considers both behaviors simultaneously, termed compositional causal reasoning (CCR): the ability to infer how causal measures compose and, equivalently, how causal quantities propagate through graphs. We instantiate a framework for the systematic evaluation of CCR for the average treatment effect and the probability of necessity and sufficiency. As proof of concept, we demonstrate the design of CCR tasks for language models in the LLama, Phi, and GPT families. On a math word problem, our framework revealed a range of taxonomically distinct error patterns. Additionally, CCR errors increased with the complexity of causal paths for all models except o1.
☆ Compositional Translation: A Novel LLM-based Approach for Low-resource Machine Translation
The ability of generative large language models (LLMs) to perform in-context learning has given rise to a large body of research into how best to prompt models for various natural language processing tasks. Machine Translation (MT) has been shown to benefit from in-context examples, in particular when they are semantically similar to the sentence to translate. In this paper, we propose a new LLM-based translation paradigm, compositional translation, to replace naive few-shot MT with similarity-based demonstrations. An LLM is used to decompose a sentence into simpler phrases, and then to translate each phrase with the help of retrieved demonstrations. Finally, the LLM is prompted to translate the initial sentence with the help of the self-generated phrase-translation pairs. Our intuition is that this approach should improve translation because these shorter phrases should be intrinsically easier to translate and easier to match with relevant examples. This is especially beneficial in low-resource scenarios, and more generally whenever the selection pool is small or out of domain. We show that compositional translation boosts LLM translation performance on a wide range of popular MT benchmarks, including FLORES 200, NTREX 128 and TICO-19. Code and outputs are available at https://github.com/ArmelRandy/compositional-translation
☆ An Empirical Study on Eliciting and Improving R1-like Reasoning Models
In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore, we show that even when a model like DeepSeek-R1-Distill-Qwen-1.5B has already achieved a high performance level, it can be further refined through RL training, reaching an accuracy of 39.33% on AIME 2024. Beyond RL training, we also explore the use of tool manipulation, finding that it significantly boosts the reasoning performance of large reasoning models. This approach achieves a remarkable accuracy of 86.67% with greedy search on AIME 2024, underscoring its effectiveness in enhancing model capabilities. We release our resources at the STILL project website: https://github.com/RUCAIBox/Slow_Thinking_with_LLMs.
comment: Technical Report on Slow Thinking with LLMs: Part III
☆ Keeping Yourself is Important in Downstream Tuning Multimodal Large Language Model
Multi-modal Large Language Models (MLLMs) integrate visual and linguistic reasoning to address complex tasks such as image captioning and visual question answering. While MLLMs demonstrate remarkable versatility, MLLMs appears limited performance on special applications. But tuning MLLMs for downstream tasks encounters two key challenges: Task-Expert Specialization, where distribution shifts between pre-training and target datasets constrain target performance, and Open-World Stabilization, where catastrophic forgetting erases the model general knowledge. In this work, we systematically review recent advancements in MLLM tuning methodologies, classifying them into three paradigms: (I) Selective Tuning, (II) Additive Tuning, and (III) Reparameterization Tuning. Furthermore, we benchmark these tuning strategies across popular MLLM architectures and diverse downstream tasks to establish standardized evaluation analysis and systematic tuning principles. Finally, we highlight several open challenges in this domain and propose future research directions. To facilitate ongoing progress in this rapidly evolving field, we provide a public repository that continuously tracks developments: https://github.com/WenkeHuang/Awesome-MLLM-Tuning.
Large Language Models in Bioinformatics: A Survey
Large Language Models (LLMs) are revolutionizing bioinformatics, enabling advanced analysis of DNA, RNA, proteins, and single-cell data. This survey provides a systematic review of recent advancements, focusing on genomic sequence modeling, RNA structure prediction, protein function inference, and single-cell transcriptomics. Meanwhile, we also discuss several key challenges, including data scarcity, computational complexity, and cross-omics integration, and explore future directions such as multimodal learning, hybrid AI models, and clinical applications. By offering a comprehensive perspective, this paper underscores the transformative potential of LLMs in driving innovations in bioinformatics and precision medicine.
☆ Generalized Interpolating Discrete Diffusion
While state-of-the-art language models achieve impressive results through next-token prediction, they have inherent limitations such as the inability to revise already generated tokens. This has prompted exploration of alternative approaches such as discrete diffusion. However, masked diffusion, which has emerged as a popular choice due to its simplicity and effectiveness, reintroduces this inability to revise words. To overcome this, we generalize masked diffusion and derive the theoretical backbone of a family of general interpolating discrete diffusion (GIDD) processes offering greater flexibility in the design of the noising processes. Leveraging a novel diffusion ELBO, we achieve compute-matched state-of-the-art performance in diffusion language modeling. Exploiting GIDD's flexibility, we explore a hybrid approach combining masking and uniform noise, leading to improved sample quality and unlocking the ability for the model to correct its own mistakes, an area where autoregressive models notoriously have struggled. Our code and models are open-source: https://github.com/dvruette/gidd/
☆ Guiding LLMs to Generate High-Fidelity and High-Quality Counterfactual Explanations for Text Classification
The need for interpretability in deep learning has driven interest in counterfactual explanations, which identify minimal changes to an instance that change a model's prediction. Current counterfactual (CF) generation methods require task-specific fine-tuning and produce low-quality text. Large Language Models (LLMs), though effective for high-quality text generation, struggle with label-flipping counterfactuals (i.e., counterfactuals that change the prediction) without fine-tuning. We introduce two simple classifier-guided approaches to support counterfactual generation by LLMs, eliminating the need for fine-tuning while preserving the strengths of LLMs. Despite their simplicity, our methods outperform state-of-the-art counterfactual generation methods and are effective across different LLMs, highlighting the benefits of guiding counterfactual generation by LLMs with classifier information. We further show that data augmentation by our generated CFs can improve a classifier's robustness. Our analysis reveals a critical issue in counterfactual generation by LLMs: LLMs rely on parametric knowledge rather than faithfully following the classifier.
☆ Quantifying patterns of punctuation in modern Chinese prose
Recent research shows that punctuation patterns in texts exhibit universal features across languages. Analysis of Western classical literature reveals that the distribution of spaces between punctuation marks aligns with a discrete Weibull distribution, typically used in survival analysis. By extending this analysis to Chinese literature represented here by three notable contemporary works, it is shown that Zipf's law applies to Chinese texts similarly to Western texts, where punctuation patterns also improve adherence to the law. Additionally, the distance distribution between punctuation marks in Chinese texts follows the Weibull model, though larger spacing is less frequent than in English translations. Sentence-ending punctuation, representing sentence length, diverges more from this pattern, reflecting greater flexibility in sentence length. This variability supports the formation of complex, multifractal sentence structures, particularly evident in Gao Xingjian's "Soul Mountain". These findings demonstrate that both Chinese and Western texts share universal punctuation and word distribution patterns, underscoring their broad applicability across languages.
☆ A Dataset for Analysing News Framing in Chinese Media
Framing is an essential device in news reporting, allowing the writer to influence public perceptions of current affairs. While there are existing automatic news framing detection datasets in various languages, none of them focus on news framing in the Chinese language which has complex character meanings and unique linguistic features. This study introduces the first Chinese News Framing dataset, to be used as either a stand-alone dataset or a supplementary resource to the SemEval-2023 task 3 dataset. We detail its creation and we run baseline experiments to highlight the need for such a dataset and create benchmarks for future research, providing results obtained through fine-tuning XLM-RoBERTa-Base and using GPT-4o in the zero-shot setting. We find that GPT-4o performs significantly worse than fine-tuned XLM-RoBERTa across all languages. For the Chinese language, we obtain an F1-micro (the performance metric for SemEval task 3, subtask 2) score of 0.719 using only samples from our Chinese News Framing dataset and a score of 0.753 when we augment the SemEval dataset with Chinese news framing samples. With positive news frame detection results, this dataset is a valuable resource for detecting news frames in the Chinese language and is a valuable supplement to the SemEval-2023 task 3 dataset.
☆ Revisiting the Othello World Model Hypothesis ICLR
Li et al. (2023) used the Othello board game as a test case for the ability of GPT-2 to induce world models, and were followed up by Nanda et al. (2023b). We briefly discuss the original experiments, expanding them to include more language models with more comprehensive probing. Specifically, we analyze sequences of Othello board states and train the model to predict the next move based on previous moves. We evaluate seven language models (GPT-2, T5, Bart, Flan-T5, Mistral, LLaMA-2, and Qwen2.5) on the Othello task and conclude that these models not only learn to play Othello, but also induce the Othello board layout. We find that all models achieve up to 99% accuracy in unsupervised grounding and exhibit high similarity in the board features they learned. This provides considerably stronger evidence for the Othello World Model Hypothesis than previous works.
comment: ICLR World Models Workshop
☆ Can Large Language Models Predict Antimicrobial Resistance Gene?
This study demonstrates that generative large language models can be utilized in a more flexible manner for DNA sequence analysis and classification tasks compared to traditional transformer encoder-based models. While recent encoder-based models such as DNABERT and Nucleotide Transformer have shown significant performance in DNA sequence classification, transformer decoder-based generative models have not yet been extensively explored in this field. This study evaluates how effectively generative Large Language Models handle DNA sequences with various labels and analyzes performance changes when additional textual information is provided. Experiments were conducted on antimicrobial resistance genes, and the results show that generative Large Language Models can offer comparable or potentially better predictions, demonstrating flexibility and accuracy when incorporating both sequence and textual information. The code and data used in this work are available at the following GitHub repository: https://github.com/biocomgit/llm4dna.
☆ Comparative Study of Zero-Shot Cross-Lingual Transfer for Bodo POS and NER Tagging Using Gemini 2.0 Flash Thinking Experimental Model
Named Entity Recognition (NER) and Part-of-Speech (POS) tagging are critical tasks for Natural Language Processing (NLP), yet their availability for low-resource languages (LRLs) like Bodo remains limited. This article presents a comparative empirical study investigating the effectiveness of Google's Gemini 2.0 Flash Thinking Experiment model for zero-shot cross-lingual transfer of POS and NER tagging to Bodo. We explore two distinct methodologies: (1) direct translation of English sentences to Bodo followed by tag transfer, and (2) prompt-based tag transfer on parallel English-Bodo sentence pairs. Both methods leverage the machine translation and cross-lingual understanding capabilities of Gemini 2.0 Flash Thinking Experiment to project English POS and NER annotations onto Bodo text in CONLL-2003 format. Our findings reveal the capabilities and limitations of each approach, demonstrating that while both methods show promise for bootstrapping Bodo NLP, prompt-based transfer exhibits superior performance, particularly for NER. We provide a detailed analysis of the results, highlighting the impact of translation quality, grammatical divergences, and the inherent challenges of zero-shot cross-lingual transfer. The article concludes by discussing future research directions, emphasizing the need for hybrid approaches, few-shot fine-tuning, and the development of dedicated Bodo NLP resources to achieve high-accuracy POS and NER tagging for this low-resource language.
comment: Submitted to SpringerNature MTAP journal. This article has not been reviewed yet. Submitting for public review!
☆ TableLoRA: Low-rank Adaptation on Table Structure Understanding for Large Language Models
Tabular data are crucial in many fields and their understanding by large language models (LLMs) under high parameter efficiency paradigm is important. However, directly applying parameter-efficient fine-tuning (PEFT) techniques to tabular tasks presents significant challenges, particularly in terms of better table serialization and the representation of two-dimensional structured information within a one-dimensional sequence. To address this, we propose TableLoRA, a module designed to improve LLMs' understanding of table structure during PEFT. It incorporates special tokens for serializing tables with special token encoder and uses 2D LoRA to encode low-rank information on cell positions. Experiments on four tabular-related datasets demonstrate that TableLoRA consistently outperforms vanilla LoRA and surpasses various table encoding methods tested in control experiments. These findings reveal that TableLoRA, as a table-specific LoRA, enhances the ability of LLMs to process tabular data effectively, especially in low-parameter settings, demonstrating its potential as a robust solution for handling table-related tasks.
☆ Shaping Shared Languages: Human and Large Language Models' Inductive Biases in Emergent Communication
Languages are shaped by the inductive biases of their users. Using a classical referential game, we investigate how artificial languages evolve when optimised for inductive biases in humans and large language models (LLMs) via Human-Human, LLM-LLM and Human-LLM experiments. We show that referentially grounded vocabularies emerge that enable reliable communication in all conditions, even when humans and LLMs collaborate. Comparisons between conditions reveal that languages optimised for LLMs subtly differ from those optimised for humans. Interestingly, interactions between humans and LLMs alleviate these differences and result in vocabularies which are more human-like than LLM-like. These findings advance our understanding of how inductive biases in LLMs play a role in the dynamic nature of human language and contribute to maintaining alignment in human and machine communication. In particular, our work underscores the need to think of new methods that include human interaction in the training processes of LLMs, and shows that using communicative success as a reward signal can be a fruitful, novel direction.
☆ More Documents, Same Length: Isolating the Challenge of Multiple Documents in RAG
Retrieval-augmented generation (RAG) provides LLMs with relevant documents. Although previous studies noted that retrieving many documents can degrade performance, they did not isolate how the quantity of documents affects performance while controlling for context length. We evaluate various language models on custom datasets derived from a multi-hop QA task. We keep the context length and position of relevant information constant while varying the number of documents, and find that increasing the document count in RAG settings poses significant challenges for LLMs. Additionally, our results indicate that processing multiple documents is a separate challenge from handling long contexts. We also make the datasets and code available: https://github.com/shaharl6000/MoreDocsSameLen .
comment: Preprint
☆ TRACT: Regression-Aware Fine-tuning Meets Chain-of-Thought Reasoning for LLM-as-a-Judge
The LLM-as-a-judge paradigm uses large language models (LLMs) for automated text evaluation, where a numerical assessment is assigned by an LLM to the input text following scoring rubrics. Existing methods for LLM-as-a-judge use cross-entropy (CE) loss for fine-tuning, which neglects the numeric nature of score prediction. Recent work addresses numerical prediction limitations of LLM fine-tuning through regression-aware fine-tuning, which, however, does not consider chain-of-thought (CoT) reasoning for score prediction. In this paper, we introduce TRACT (Two-stage Regression-Aware fine-tuning with CoT), a method combining CoT reasoning with regression-aware training. TRACT consists of two stages: first, seed LLM is fine-tuned to generate CoTs, which serve as supervision for the second stage fine-tuning. The training objective of TRACT combines the CE loss for learning the CoT reasoning capabilities, and the regression-aware loss for the score prediction. Experiments across four LLM-as-a-judge datasets and two LLMs show that TRACT significantly outperforms existing methods. Extensive ablation studies validate the importance of each component in TRACT.
comment: Codes and models are available at https://github.com/d223302/TRACT
☆ Dedicated Feedback and Edit Models Empower Inference-Time Scaling for Open-Ended General-Domain Tasks
Inference-Time Scaling has been critical to the success of recent models such as OpenAI o1 and DeepSeek R1. However, many techniques used to train models for inference-time scaling require tasks to have answers that can be verified, limiting their application to domains such as math, coding and logical reasoning. We take inspiration from how humans make first attempts, ask for detailed feedback from others and make improvements based on such feedback across a wide spectrum of open-ended endeavors. To this end, we collect data for and train dedicated Feedback and Edit Models that are capable of performing inference-time scaling for open-ended general-domain tasks. In our setup, one model generates an initial response, which are given feedback by a second model, that are then used by a third model to edit the response. We show that performance on Arena Hard, a benchmark strongly predictive of Chatbot Arena Elo can be boosted by scaling the number of initial response drafts, effective feedback and edited responses. When scaled optimally, our setup based on 70B models from the Llama 3 family can reach SoTA performance on Arena Hard at 92.7 as of 5 Mar 2025, surpassing OpenAI o1-preview-2024-09-12 with 90.4 and DeepSeek R1 with 92.3.
comment: 22 pages, 2 figures
☆ Assumed Identities: Quantifying Gender Bias in Machine Translation of Ambiguous Occupational Terms
Machine Translation (MT) systems frequently encounter ambiguous scenarios where they must assign gender to certain occupations when translating without explicit guidance or contextual cues. While individual translations in such cases may not be inherently biased, systematic patterns-such as the repeated association of certain professions with specific genders-can emerge, reflecting and perpetuating societal stereotypes. This ambiguity challenges traditional instance-level single-answer evaluation approaches, as no single gold standard translation exists. To address this, we propose an approach that evaluates gender bias through aggregated model responses. Specifically, we introduce a methodology to detect gender imbalances between source texts and translations, a benchmarking dataset with ambiguous English inputs, and probability-based metrics to quantify a model's divergence from normative standards or reference distributions.
☆ Lost in Literalism: How Supervised Training Shapes Translationese in LLMs
Large language models (LLMs) have achieved remarkable success in machine translation, demonstrating impressive performance across diverse languages. However, translationese, characterized by overly literal and unnatural translations, remains a persistent challenge in LLM-based translation systems. Despite their pre-training on vast corpora of natural utterances, LLMs exhibit translationese errors and generate unexpected unnatural translations, stemming from biases introduced during supervised fine-tuning (SFT). In this work, we systematically evaluate the prevalence of translationese in LLM-generated translations and investigate its roots during supervised training. We introduce methods to mitigate these biases, including polishing golden references and filtering unnatural training instances. Empirical evaluations demonstrate that these approaches significantly reduce translationese while improving translation naturalness, validated by human evaluations and automatic metrics. Our findings highlight the need for training-aware adjustments to optimize LLM translation outputs, paving the way for more fluent and target-language-consistent translations. We release the data and code at https://github.com/yafuly/LLM_Translationese.
comment: 19 pages;
☆ Exploring the Multilingual NLG Evaluation Abilities of LLM-Based Evaluators
Previous research has shown that LLMs have potential in multilingual NLG evaluation tasks. However, existing research has not fully explored the differences in the evaluation capabilities of LLMs across different languages. To this end, this study provides a comprehensive analysis of the multilingual evaluation performance of 10 recent LLMs, spanning high-resource and low-resource languages through correlation analysis, perturbation attacks, and fine-tuning. We found that 1) excluding the reference answer from the prompt and using large-parameter LLM-based evaluators leads to better performance across various languages; 2) most LLM-based evaluators show a higher correlation with human judgments in high-resource languages than in low-resource languages; 3) in the languages where they are most sensitive to such attacks, they also tend to exhibit the highest correlation with human judgments; and 4) fine-tuning with data from a particular language yields a broadly consistent enhancement in the model's evaluation performance across diverse languages. Our findings highlight the imbalance in LLMs'evaluation capabilities across different languages and suggest that low-resource language scenarios deserve more attention.
☆ Layer-Specific Scaling of Positional Encodings for Superior Long-Context Modeling
Although large language models (LLMs) have achieved significant progress in handling long-context inputs, they still suffer from the ``lost-in-the-middle'' problem, where crucial information in the middle of the context is often underrepresented or lost. Our extensive experiments reveal that this issue may arise from the rapid long-term decay in Rotary Position Embedding (RoPE). To address this problem, we propose a layer-specific positional encoding scaling method that assigns distinct scaling factors to each layer, slowing down the decay rate caused by RoPE to make the model pay more attention to the middle context. A specially designed genetic algorithm is employed to efficiently select the optimal scaling factors for each layer by incorporating Bezier curves to reduce the search space. Through comprehensive experimentation, we demonstrate that our method significantly alleviates the ``lost-in-the-middle'' problem. Our approach results in an average accuracy improvement of up to 20% on the Key-Value Retrieval dataset. Furthermore, we show that layer-specific interpolation, as opposed to uniform interpolation across all layers, enhances the model's extrapolation capabilities when combined with PI and Dynamic-NTK positional encoding schemes.
☆ Adding Alignment Control to Language Models
Post-training alignment has increasingly become a crucial factor in enhancing the usability of language models (LMs). However, the strength of alignment varies depending on individual preferences. This paper proposes a method to incorporate alignment control into a single model, referred to as CLM. This approach adds one identity layer preceding the initial layers and performs preference learning only on this layer to map unaligned input token embeddings into the aligned space. Experimental results demonstrate that this efficient fine-tuning method performs comparable to full fine-tuning. During inference, the input embeddings are processed through the aligned and unaligned layers, which are then merged through the interpolation coefficient. By controlling this parameter, the alignment exhibits a clear interpolation and extrapolation phenomenon.
☆ In-depth Analysis of Graph-based RAG in a Unified Framework
Graph-based Retrieval-Augmented Generation (RAG) has proven effective in integrating external knowledge into large language models (LLMs), improving their factual accuracy, adaptability, interpretability, and trustworthiness. A number of graph-based RAG methods have been proposed in the literature. However, these methods have not been systematically and comprehensively compared under the same experimental settings. In this paper, we first summarize a unified framework to incorporate all graph-based RAG methods from a high-level perspective. We then extensively compare representative graph-based RAG methods over a range of questing-answering (QA) datasets -- from specific questions to abstract questions -- and examine the effectiveness of all methods, providing a thorough analysis of graph-based RAG approaches. As a byproduct of our experimental analysis, we are also able to identify new variants of the graph-based RAG methods over specific QA and abstract QA tasks respectively, by combining existing techniques, which outperform the state-of-the-art methods. Finally, based on these findings, we offer promising research opportunities. We believe that a deeper understanding of the behavior of existing methods can provide new valuable insights for future research.
☆ Solving Word-Sense Disambiguation and Word-Sense Induction with Dictionary Examples
Many less-resourced languages struggle with a lack of large, task-specific datasets that are required for solving relevant tasks with modern transformer-based large language models (LLMs). On the other hand, many linguistic resources, such as dictionaries, are rarely used in this context despite their large information contents. We show how LLMs can be used to extend existing language resources in less-resourced languages for two important tasks: word-sense disambiguation (WSD) and word-sense induction (WSI). We approach the two tasks through the related but much more accessible word-in-context (WiC) task where, given a pair of sentences and a target word, a classification model is tasked with predicting whether the sense of a given word differs between sentences. We demonstrate that a well-trained model for this task can distinguish between different word senses and can be adapted to solve the WSD and WSI tasks. The advantage of using the WiC task, instead of directly predicting senses, is that the WiC task does not need pre-constructed sense inventories with a sufficient number of examples for each sense, which are rarely available in less-resourced languages. We show that sentence pairs for the WiC task can be successfully generated from dictionary examples using LLMs. The resulting prediction models outperform existing models on WiC, WSD, and WSI tasks. We demonstrate our methodology on the Slovene language, where a monolingual dictionary is available, but word-sense resources are tiny.
comment: 12 pages, 1 figure
☆ Computational Law: Datasets, Benchmarks, and Ontologies
Recent developments in computer science and artificial intelligence have also contributed to the legal domain, as revealed by the number and range of related publications and applications. Machine and deep learning models require considerable amount of domain-specific data for training and comparison purposes, in order to attain high-performance in the legal domain. Additionally, semantic resources such as ontologies are valuable for building large-scale computational legal systems, in addition to ensuring interoperability of such systems. Considering these aspects, we present an up-to-date review of the literature on datasets, benchmarks, and ontologies proposed for computational law. We believe that this comprehensive and recent review will help researchers and practitioners when developing and testing approaches and systems for computational law.
☆ Dual-Class Prompt Generation: Enhancing Indonesian Gender-Based Hate Speech Detection through Data Augmentation
Detecting gender-based hate speech in Indonesian social media remains challenging due to limited labeled datasets. While binary hate speech classification has advanced, a more granular category like gender-targeted hate speech is understudied because of class imbalance issues. This paper addresses this gap by comparing three data augmentation techniques for Indonesian gender-based hate speech detection. We evaluate backtranslation, single-class prompt generation (using only hate speech examples), and our proposed dual-class prompt generation (using both hate speech and non-hate speech examples). Experiments show all augmentation methods improve classification performance, with our dual-class approach achieving the best results (88.5% accuracy, 88.1% F1-score using Random Forest). Semantic similarity analysis reveals dual-class prompt generation produces the most novel content, while T-SNE visualizations confirm these samples occupy distinct feature space regions while maintaining class characteristics. Our findings suggest that incorporating examples from both classes helps language models generate more diverse yet representative samples, effectively addressing limited data challenges in specialized hate speech detection.
comment: Accepted to the 8th World Conference on Computing and Communication Technologies (WCCCT 2025)
☆ On Fact and Frequency: LLM Responses to Misinformation Expressed with Uncertainty
We study LLM judgments of misinformation expressed with uncertainty. Our experiments study the response of three widely used LLMs (GPT-4o, LlaMA3, DeepSeek-v2) to misinformation propositions that have been verified false and then are transformed into uncertain statements according to an uncertainty typology. Our results show that after transformation, LLMs change their factchecking classification from false to not-false in 25% of the cases. Analysis reveals that the change cannot be explained by predictors to which humans are expected to be sensitive, i.e., modality, linguistic cues, or argumentation strategy. The exception is doxastic transformations, which use linguistic cue phrases such as "It is believed ...".To gain further insight, we prompt the LLM to make another judgment about the transformed misinformation statements that is not related to truth value. Specifically, we study LLM estimates of the frequency with which people make the uncertain statement. We find a small but significant correlation between judgment of fact and estimation of frequency.
comment: 4 pages, 1 figure, 3 tables, conference
☆ DiffPO: Diffusion-styled Preference Optimization for Efficient Inference-Time Alignment of Large Language Models
Inference-time alignment provides an efficient alternative for aligning LLMs with humans. However, these approaches still face challenges, such as limited scalability due to policy-specific value functions and latency during the inference phase. In this paper, we propose a novel approach, Diffusion-styled Preference Optimization (\model), which provides an efficient and policy-agnostic solution for aligning LLMs with humans. By directly performing alignment at sentence level, \model~avoids the time latency associated with token-level generation. Designed as a plug-and-play module, \model~can be seamlessly integrated with various base models to enhance their alignment. Extensive experiments on AlpacaEval 2, MT-bench, and HH-RLHF demonstrate that \model~achieves superior alignment performance across various settings, achieving a favorable trade-off between alignment quality and inference-time latency. Furthermore, \model~demonstrates model-agnostic scalability, significantly improving the performance of large models such as Llama-3-70B.
☆ Tgea: An error-annotated dataset and benchmark tasks for text generation from pretrained language models ACL 2021
In order to deeply understand the capability of pretrained language models in text generation and conduct a diagnostic evaluation, we propose TGEA, an error-annotated dataset with multiple benchmark tasks for text generation from pretrained language models (PLMs). We use carefully selected prompt words to guide GPT-2 to generate candidate sentences, from which we select 47K for error annotation. Crowdsourced workers manually check each of these sentences and detect 12k erroneous sentences. We create an error taxonomy to cover 24 types of errors occurring in these erroneous sentences according to the nature of errors with respect to linguistics and knowledge (eg, common sense). For each erroneous span in PLM-generated sentences, we also detect another span that is closely associated with it. Each error is hence manually labeled with comprehensive annotations, including the span of the error, the associated span, minimal correction to the error, the type of the error, and rationale behind the error. Apart from the fully annotated dataset, we also present a detailed description of the data collection procedure, statistics and analysis of the dataset. This is the first dataset with comprehensive annotations for PLM-generated texts, which facilitates the diagnostic evaluation of PLM-based text generation. Furthermore, we use TGEA as a benchmark dataset and propose a series of automatic diagnosis tasks, including error detection, error type classification, associated span detection, error rationale generation, to further promote future study on the automatic error detection and correction on texts generated by pretrained language models.
comment: ACL 2021
☆ FuseChat-3.0: Preference Optimization Meets Heterogeneous Model Fusion
We introduce FuseChat-3.0, a suite of large language models (LLMs) developed by integrating the strengths of heterogeneous source LLMs into more compact target LLMs. Our source models include the powerful Gemma-2-27B-it, Mistral-Large-Instruct-2407, Qwen-2.5-72B-Instruct, and Llama-3.1-70B-Instruct. For target models, we focus on three widely-used smaller variants-Llama-3.1-8B-Instruct, Gemma-2-9B-it, and Qwen-2.5-7B-Instruct-along with two ultra-compact options, Llama-3.2-3B-Instruct and Llama-3.2-1B-Instruct. To leverage the diverse capabilities of these source models, we develop a specialized data construction protocol tailored to various tasks and domains. The FuseChat-3.0 training pipeline consists of two key stages: (1) supervised fine-tuning (SFT) to align the target and source model distributions, and (2) Direct Preference Optimization (DPO) to apply preferences from multiple source LLMs to fine-tune the target model. The resulting FuseChat-3.0 models exhibit significant performance gains across tasks such as instruction following, general knowledge, mathematics, and coding. As illustrated in Figure 1, using Llama-3.1-8B-Instruct as the target model, our fusion approach achieves an average improvement of 6.8 points across 14 benchmarks. Moreover, it demonstrates remarkable gains of 37.1 points and 30.1 points on the instruction-following benchmarks AlpacaEval-2 and Arena-Hard, respectively. Our code, models, and datasets are available at https://github.com/SLIT-AI/FuseChat-3.0.
comment: Technical report
☆ Knowledge-Decoupled Synergetic Learning: An MLLM based Collaborative Approach to Few-shot Multimodal Dialogue Intention Recognition
Few-shot multimodal dialogue intention recognition is a critical challenge in the e-commerce domainn. Previous methods have primarily enhanced model classification capabilities through post-training techniques. However, our analysis reveals that training for few-shot multimodal dialogue intention recognition involves two interconnected tasks, leading to a seesaw effect in multi-task learning. This phenomenon is attributed to knowledge interference stemming from the superposition of weight matrix updates during the training process. To address these challenges, we propose Knowledge-Decoupled Synergetic Learning (KDSL), which mitigates these issues by utilizing smaller models to transform knowledge into interpretable rules, while applying the post-training of larger models. By facilitating collaboration between the large and small multimodal large language models for prediction, our approach demonstrates significant improvements. Notably, we achieve outstanding results on two real Taobao datasets, with enhancements of 6.37\% and 6.28\% in online weighted F1 scores compared to the state-of-the-art method, thereby validating the efficacy of our framework.
☆ Measuring temporal effects of agent knowledge by date-controlled tool use
Temporal progression is an integral part of knowledge accumulation and update. Web search is frequently adopted as grounding for agent knowledge, yet its inappropriate configuration affects the quality of agent responses. Here, we construct a tool-based out-of-sample testing framework to measure the knowledge variability of large language model (LLM) agents from distinct date-controlled tools (DCTs). We demonstrate the temporal effects of an LLM agent as a writing assistant, which can use web search to help complete scientific publication abstracts. We show that temporal effects of the search engine translates into tool-dependent agent performance but can be alleviated with base model choice and explicit reasoning instructions such as chain-of-thought prompting. Our results indicate that agent evaluation should take a dynamical view and account for the temporal influence of tools and the updates of external resources.
comment: comments welcome
☆ Large-Scale AI in Telecom: Charting the Roadmap for Innovation, Scalability, and Enhanced Digital Experiences
This white paper discusses the role of large-scale AI in the telecommunications industry, with a specific focus on the potential of generative AI to revolutionize network functions and user experiences, especially in the context of 6G systems. It highlights the development and deployment of Large Telecom Models (LTMs), which are tailored AI models designed to address the complex challenges faced by modern telecom networks. The paper covers a wide range of topics, from the architecture and deployment strategies of LTMs to their applications in network management, resource allocation, and optimization. It also explores the regulatory, ethical, and standardization considerations for LTMs, offering insights into their future integration into telecom infrastructure. The goal is to provide a comprehensive roadmap for the adoption of LTMs to enhance scalability, performance, and user-centric innovation in telecom networks.
☆ TIMER: Temporal Instruction Modeling and Evaluation for Longitudinal Clinical Records
Large language models (LLMs) have emerged as promising tools for assisting in medical tasks, yet processing Electronic Health Records (EHRs) presents unique challenges due to their longitudinal nature. While LLMs' capabilities to perform medical tasks continue to improve, their ability to reason over temporal dependencies across multiple patient visits and time frames remains unexplored. We introduce TIMER (Temporal Instruction Modeling and Evaluation for Longitudinal Clinical Records), a framework that incorporate instruction-response pairs grounding to different parts of a patient's record as a critical dimension in both instruction evaluation and tuning for longitudinal clinical records. We develop TIMER-Bench, the first time-aware benchmark that evaluates temporal reasoning capabilities over longitudinal EHRs, as well as TIMER-Instruct, an instruction-tuning methodology for LLMs to learn reasoning over time. We demonstrate that models fine-tuned with TIMER-Instruct improve performance by 7.3% on human-generated benchmarks and 9.2% on TIMER-Bench, indicating that temporal instruction-tuning improves model performance for reasoning over EHR.
comment: Preprint
☆ BPQA Dataset: Evaluating How Well Language Models Leverage Blood Pressures to Answer Biomedical Questions
Clinical measurements such as blood pressures and respiration rates are critical in diagnosing and monitoring patient outcomes. It is an important component of biomedical data, which can be used to train transformer-based language models (LMs) for improving healthcare delivery. It is, however, unclear whether LMs can effectively interpret and use clinical measurements. We investigate two questions: First, can LMs effectively leverage clinical measurements to answer related medical questions? Second, how to enhance an LM's performance on medical question-answering (QA) tasks that involve measurements? We performed a case study on blood pressure readings (BPs), a vital sign routinely monitored by medical professionals. We evaluated the performance of four LMs: BERT, BioBERT, MedAlpaca, and GPT-3.5, on our newly developed dataset, BPQA (Blood Pressure Question Answering). BPQA contains $100$ medical QA pairs that were verified by medical students and designed to rely on BPs . We found that GPT-3.5 and MedAlpaca (larger and medium sized LMs) benefit more from the inclusion of BPs than BERT and BioBERT (small sized LMs). Further, augmenting measurements with labels improves the performance of BioBERT and Medalpaca (domain specific LMs), suggesting that retrieval may be useful for improving domain-specific LMs.
comment: 9 pages
☆ Ticktack : Long Span Temporal Alignment of Large Language Models Leveraging Sexagenary Cycle Time Expression
Large language models (LLMs) suffer from temporal misalignment issues especially across long span of time. The issue arises from knowing that LLMs are trained on large amounts of data where temporal information is rather sparse over long times, such as thousands of years, resulting in insufficient learning or catastrophic forgetting by the LLMs. This paper proposes a methodology named "Ticktack" for addressing the LLM's long-time span misalignment in a yearly setting. Specifically, we first propose to utilize the sexagenary year expression instead of the Gregorian year expression employed by LLMs, achieving a more uniform distribution in yearly granularity. Then, we employ polar coordinates to model the sexagenary cycle of 60 terms and the year order within each term, with additional temporal encoding to ensure LLMs understand them. Finally, we present a temporal representational alignment approach for post-training LLMs that effectively distinguishes time points with relevant knowledge, hence improving performance on time-related tasks, particularly over a long period. We also create a long time span benchmark for evaluation. Experimental results prove the effectiveness of our proposal.
☆ Dynamic Benchmarking of Reasoning Capabilities in Code Large Language Models Under Data Contamination
The rapid evolution of code largelanguage models underscores the need for effective and transparent benchmarking of their reasoning capabilities. However, the current benchmarking approach heavily depends on publicly available, human-created datasets. The widespread use of these fixed benchmark datasets makes the benchmarking process to be static and thus particularly susceptible to data contamination, an unavoidable consequence of the extensive data collection processes used to train Code LLMs. Existing approaches that address data contamination often suffer from human effort limitations and imbalanced problem complexity. To tackle these challenges, we propose \tool, a novel benchmarking suite for evaluating Code LLMs under potential data contamination. Given a seed programming problem, \tool employs multiple agents to extract and modify the context without altering the core logic, generating semantically equivalent variations. We introduce a dynamic data generation methods and conduct empirical studies on two seed datasets across 21 Code LLMs. Results show that \tool effectively benchmarks reasoning capabilities under contamination risks while generating diverse problem sets to ensure consistent and reliable evaluations.
comment: https://codekaleidoscope.github.io/dycodeeval.html
☆ HEISIR: Hierarchical Expansion of Inverted Semantic Indexing for Training-free Retrieval of Conversational Data using LLMs NAACL 2025
The growth of conversational AI services has increased demand for effective information retrieval from dialogue data. However, existing methods often face challenges in capturing semantic intent or require extensive labeling and fine-tuning. This paper introduces HEISIR (Hierarchical Expansion of Inverted Semantic Indexing for Retrieval), a novel framework that enhances semantic understanding in conversational data retrieval through optimized data ingestion, eliminating the need for resource-intensive labeling or model adaptation. HEISIR implements a two-step process: (1) Hierarchical Triplets Formulation and (2) Adjunct Augmentation, creating semantic indices consisting of Subject-Verb-Object-Adjunct (SVOA) quadruplets. This structured representation effectively captures the underlying semantic information from dialogue content. HEISIR achieves high retrieval performance while maintaining low latency during the actual retrieval process. Our experimental results demonstrate that HEISIR outperforms fine-tuned models across various embedding types and language models. Beyond improving retrieval capabilities, HEISIR also offers opportunities for intent and topic analysis in conversational data, providing a versatile solution for dialogue systems.
comment: Accepted by NAACL 2025 (Findings)
☆ Biological Sequence with Language Model Prompting: A Survey
Large Language models (LLMs) have emerged as powerful tools for addressing challenges across diverse domains. Notably, recent studies have demonstrated that large language models significantly enhance the efficiency of biomolecular analysis and synthesis, attracting widespread attention from academics and medicine. In this paper, we systematically investigate the application of prompt-based methods with LLMs to biological sequences, including DNA, RNA, proteins, and drug discovery tasks. Specifically, we focus on how prompt engineering enables LLMs to tackle domain-specific problems, such as promoter sequence prediction, protein structure modeling, and drug-target binding affinity prediction, often with limited labeled data. Furthermore, our discussion highlights the transformative potential of prompting in bioinformatics while addressing key challenges such as data scarcity, multimodal fusion, and computational resource limitations. Our aim is for this paper to function both as a foundational primer for newcomers and a catalyst for continued innovation within this dynamic field of study.
☆ Uncovering Gaps in How Humans and LLMs Interpret Subjective Language ICLR 2025
Humans often rely on subjective natural language to direct language models (LLMs); for example, users might instruct the LLM to write an enthusiastic blogpost, while developers might train models to be helpful and harmless using LLM-based edits. The LLM's operational semantics of such subjective phrases -- how it adjusts its behavior when each phrase is included in the prompt -- thus dictates how aligned it is with human intent. In this work, we uncover instances of misalignment between LLMs' actual operational semantics and what humans expect. Our method, TED (thesaurus error detector), first constructs a thesaurus that captures whether two phrases have similar operational semantics according to the LLM. It then elicits failures by unearthing disagreements between this thesaurus and a human-constructed reference. TED routinely produces surprising instances of misalignment; for example, Mistral 7B Instruct produces more harassing outputs when it edits text to be witty, and Llama 3 8B Instruct produces dishonest articles when instructed to make the articles enthusiastic. Our results demonstrate that humans can uncover unexpected LLM behavior by scrutinizing relationships between abstract concepts, without supervising outputs directly.
comment: Published at ICLR 2025
LLMs Can Generate a Better Answer by Aggregating Their Own Responses
Large Language Models (LLMs) have shown remarkable capabilities across tasks, yet they often require additional prompting techniques when facing complex problems. While approaches like self-correction and response selection have emerged as popular solutions, recent studies have shown these methods perform poorly when relying on the LLM itself to provide feedback or selection criteria. We argue this limitation stems from the fact that common LLM post-training procedures lack explicit supervision for discriminative judgment tasks. In this paper, we propose Generative Self-Aggregation (GSA), a novel prompting method that improves answer quality without requiring the model's discriminative capabilities. GSA first samples multiple diverse responses from the LLM, then aggregates them to obtain an improved solution. Unlike previous approaches, our method does not require the LLM to correct errors or compare response quality; instead, it leverages the model's generative abilities to synthesize a new response based on the context of multiple samples. While GSA shares similarities with the self-consistency (SC) approach for response aggregation, SC requires specific verifiable tokens to enable majority voting. In contrast, our approach is more general and can be applied to open-ended tasks. Empirical evaluation demonstrates that GSA effectively improves response quality across various tasks, including mathematical reasoning, knowledge-based problems, and open-ended generation tasks such as code synthesis and conversational responses.
☆ Disparities in LLM Reasoning Accuracy and Explanations: A Case Study on African American English
Large Language Models (LLMs) have demonstrated remarkable capabilities in reasoning tasks, leading to their widespread deployment. However, recent studies have highlighted concerning biases in these models, particularly in their handling of dialectal variations like African American English (AAE). In this work, we systematically investigate dialectal disparities in LLM reasoning tasks. We develop an experimental framework comparing LLM performance given Standard American English (SAE) and AAE prompts, combining LLM-based dialect conversion with established linguistic analyses. We find that LLMs consistently produce less accurate responses and simpler reasoning chains and explanations for AAE inputs compared to equivalent SAE questions, with disparities most pronounced in social science and humanities domains. These findings highlight systematic differences in how LLMs process and reason about different language varieties, raising important questions about the development and deployment of these systems in our multilingual and multidialectal world. Our code repository is publicly available at https://github.com/Runtaozhou/dialect_bias_eval.
comment: ARR Under Review, First two authors contribute equally
☆ Chart-HQA: A Benchmark for Hypothetical Question Answering in Charts
Multimodal Large Language Models (MLLMs) have garnered significant attention for their strong visual-semantic understanding. Most existing chart benchmarks evaluate MLLMs' ability to parse information from charts to answer questions.However, they overlook the inherent output biases of MLLMs, where models rely on their parametric memory to answer questions rather than genuinely understanding the chart content. To address this limitation, we introduce a novel Chart Hypothetical Question Answering (HQA) task, which imposes assumptions on the same question to compel models to engage in counterfactual reasoning based on the chart content. Furthermore, we introduce HAI, a human-AI interactive data synthesis approach that leverages the efficient text-editing capabilities of LLMs alongside human expert knowledge to generate diverse and high-quality HQA data at a low cost. Using HAI, we construct Chart-HQA, a challenging benchmark synthesized from publicly available data sources. Evaluation results on 18 MLLMs of varying model sizes reveal that current models face significant generalization challenges and exhibit imbalanced reasoning performance on the HQA task.
comment: Under review
☆ PP-DocBee: Improving Multimodal Document Understanding Through a Bag of Tricks
With the rapid advancement of digitalization, various document images are being applied more extensively in production and daily life, and there is an increasingly urgent need for fast and accurate parsing of the content in document images. Therefore, this report presents PP-DocBee, a novel multimodal large language model designed for end-to-end document image understanding. First, we develop a data synthesis strategy tailored to document scenarios in which we build a diverse dataset to improve the model generalization. Then, we apply a few training techniques, including dynamic proportional sampling, data preprocessing, and OCR postprocessing strategies. Extensive evaluations demonstrate the superior performance of PP-DocBee, achieving state-of-the-art results on English document understanding benchmarks and even outperforming existing open source and commercial models in Chinese document understanding. The source code and pre-trained models are publicly available at \href{https://github.com/PaddlePaddle/PaddleMIX}{https://github.com/PaddlePaddle/PaddleMIX}.
☆ Uncovering inequalities in new knowledge learning by large language models across different languages
As large language models (LLMs) gradually become integral tools for problem solving in daily life worldwide, understanding linguistic inequality is becoming increasingly important. Existing research has primarily focused on static analyses that assess the disparities in the existing knowledge and capabilities of LLMs across languages. However, LLMs are continuously evolving, acquiring new knowledge to generate up-to-date, domain-specific responses. Investigating linguistic inequalities within this dynamic process is, therefore, also essential. In this paper, we explore inequalities in new knowledge learning by LLMs across different languages and four key dimensions: effectiveness, transferability, prioritization, and robustness. Through extensive experiments under two settings (in-context learning and fine-tuning) using both proprietary and open-source models, we demonstrate that low-resource languages consistently face disadvantages across all four dimensions. By shedding light on these disparities, we aim to raise awareness of linguistic inequalities in LLMs' new knowledge learning, fostering the development of more inclusive and equitable future LLMs.
☆ Robust Data Watermarking in Language Models by Injecting Fictitious Knowledge
Data watermarking in language models injects traceable signals, such as specific token sequences or stylistic patterns, into copyrighted text, allowing copyright holders to track and verify training data ownership. Previous data watermarking techniques primarily focus on effective memorization after pretraining, while overlooking challenges that arise in other stages of the LLM pipeline, such as the risk of watermark filtering during data preprocessing, or potential forgetting through post-training, or verification difficulties due to API-only access. We propose a novel data watermarking approach that injects coherent and plausible yet fictitious knowledge into training data using generated passages describing a fictitious entity and its associated attributes. Our watermarks are designed to be memorized by the LLM through seamlessly integrating in its training data, making them harder to detect lexically during preprocessing.We demonstrate that our watermarks can be effectively memorized by LLMs, and that increasing our watermarks' density, length, and diversity of attributes strengthens their memorization. We further show that our watermarks remain robust throughout LLM development, maintaining their effectiveness after continual pretraining and supervised finetuning. Finally, we show that our data watermarks can be evaluated even under API-only access via question answering.
☆ Benchmarking Large Language Models on Multiple Tasks in Bioinformatics NLP with Prompting
Large language models (LLMs) have become important tools in solving biological problems, offering improvements in accuracy and adaptability over conventional methods. Several benchmarks have been proposed to evaluate the performance of these LLMs. However, current benchmarks can hardly evaluate the performance of these models across diverse tasks effectively. In this paper, we introduce a comprehensive prompting-based benchmarking framework, termed Bio-benchmark, which includes 30 key bioinformatics tasks covering areas such as proteins, RNA, drugs, electronic health records, and traditional Chinese medicine. Using this benchmark, we evaluate six mainstream LLMs, including GPT-4o and Llama-3.1-70b, etc., using 0-shot and few-shot Chain-of-Thought (CoT) settings without fine-tuning to reveal their intrinsic capabilities. To improve the efficiency of our evaluations, we demonstrate BioFinder, a new tool for extracting answers from LLM responses, which increases extraction accuracy by round 30% compared to existing methods. Our benchmark results show the biological tasks suitable for current LLMs and identify specific areas requiring enhancement. Furthermore, we propose targeted prompt engineering strategies for optimizing LLM performance in these contexts. Based on these findings, we provide recommendations for the development of more robust LLMs tailored for various biological applications. This work offers a comprehensive evaluation framework and robust tools to support the application of LLMs in bioinformatics.
☆ RetinalGPT: A Retinal Clinical Preference Conversational Assistant Powered by Large Vision-Language Models
Recently, Multimodal Large Language Models (MLLMs) have gained significant attention for their remarkable ability to process and analyze non-textual data, such as images, videos, and audio. Notably, several adaptations of general-domain MLLMs to the medical field have been explored, including LLaVA-Med. However, these medical adaptations remain insufficiently advanced in understanding and interpreting retinal images. In contrast, medical experts emphasize the importance of quantitative analyses for disease detection and interpretation. This underscores a gap between general-domain and medical-domain MLLMs: while general-domain MLLMs excel in broad applications, they lack the specialized knowledge necessary for precise diagnostic and interpretative tasks in the medical field. To address these challenges, we introduce \textit{RetinalGPT}, a multimodal conversational assistant for clinically preferred quantitative analysis of retinal images. Specifically, we achieve this by compiling a large retinal image dataset, developing a novel data pipeline, and employing customized visual instruction tuning to enhance both retinal analysis and enrich medical knowledge. In particular, RetinalGPT outperforms MLLM in the generic domain by a large margin in the diagnosis of retinal diseases in 8 benchmark retinal datasets. Beyond disease diagnosis, RetinalGPT features quantitative analyses and lesion localization, representing a pioneering step in leveraging LLMs for an interpretable and end-to-end clinical research framework. The code is available at https://github.com/Retinal-Research/RetinalGPT
☆ Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities
Understanding and reasoning over non-speech sounds and music are crucial for both humans and AI agents to interact effectively with their environments. In this paper, we introduce Audio Flamingo 2 (AF2), an Audio-Language Model (ALM) with advanced audio understanding and reasoning capabilities. AF2 leverages (i) a custom CLAP model, (ii) synthetic Audio QA data for fine-grained audio reasoning, and (iii) a multi-stage curriculum learning strategy. AF2 achieves state-of-the-art performance with only a 3B parameter small language model, surpassing large open-source and proprietary models across over 20 benchmarks. Next, for the first time, we extend audio understanding to long audio segments (30 secs to 5 mins) and propose LongAudio, a large and novel dataset for training ALMs on long audio captioning and question-answering tasks. Fine-tuning AF2 on LongAudio leads to exceptional performance on our proposed LongAudioBench, an expert annotated benchmark for evaluating ALMs on long audio understanding capabilities. We conduct extensive ablation studies to confirm the efficacy of our approach. Project Website: https://research.nvidia.com/labs/adlr/AF2/.
☆ ReasonGraph: Visualisation of Reasoning Paths
Large Language Models (LLMs) reasoning processes are challenging to analyze due to their complexity and the lack of organized visualization tools. We present ReasonGraph, a web-based platform for visualizing and analyzing LLM reasoning processes. It supports both sequential and tree-based reasoning methods while integrating with major LLM providers and over fifty state-of-the-art models. ReasonGraph incorporates an intuitive UI with meta reasoning method selection, configurable visualization parameters, and a modular framework that facilitates efficient extension. Our evaluation shows high parsing reliability, efficient processing, and strong usability across various downstream applications. By providing a unified visualization framework, ReasonGraph reduces cognitive load in analyzing complex reasoning paths, improves error detection in logical processes, and enables more effective development of LLM-based applications. The platform is open-source, promoting accessibility and reproducibility in LLM reasoning analysis.
☆ A Unified Framework with Novel Metrics for Evaluating the Effectiveness of XAI Techniques in LLMs
The increasing complexity of LLMs presents significant challenges to their transparency and interpretability, necessitating the use of eXplainable AI (XAI) techniques to enhance trustworthiness and usability. This study introduces a comprehensive evaluation framework with four novel metrics for assessing the effectiveness of five XAI techniques across five LLMs and two downstream tasks. We apply this framework to evaluate several XAI techniques LIME, SHAP, Integrated Gradients, Layer-wise Relevance Propagation (LRP), and Attention Mechanism Visualization (AMV) using the IMDB Movie Reviews and Tweet Sentiment Extraction datasets. The evaluation focuses on four key metrics: Human-reasoning Agreement (HA), Robustness, Consistency, and Contrastivity. Our results show that LIME consistently achieves high scores across multiple LLMs and evaluation metrics, while AMV demonstrates superior Robustness and near-perfect Consistency. LRP excels in Contrastivity, particularly with more complex models. Our findings provide valuable insights into the strengths and limitations of different XAI methods, offering guidance for developing and selecting appropriate XAI techniques for LLMs.
comment: arXiv admin note: substantial text overlap with arXiv:2501.15374
☆ Dynamic-KGQA: A Scalable Framework for Generating Adaptive Question Answering Datasets
As question answering (QA) systems advance alongside the rapid evolution of foundation models, the need for robust, adaptable, and large-scale evaluation benchmarks becomes increasingly critical. Traditional QA benchmarks are often static and publicly available, making them susceptible to data contamination and memorization by large language models (LLMs). Consequently, static benchmarks may overestimate model generalization and hinder a reliable assessment of real-world performance. In this work, we introduce Dynamic-KGQA, a scalable framework for generating adaptive QA datasets from knowledge graphs (KGs), designed to mitigate memorization risks while maintaining statistical consistency across iterations. Unlike fixed benchmarks, Dynamic-KGQA generates a new dataset variant on every run while preserving the underlying distribution, enabling fair and reproducible evaluations. Furthermore, our framework provides fine-grained control over dataset characteristics, supporting domain-specific and topic-focused QA dataset generation. Additionally, Dynamic-KGQA produces compact, semantically coherent subgraphs that facilitate both training and evaluation of KGQA models, enhancing their ability to leverage structured knowledge effectively. To align with existing evaluation protocols, we also provide static large-scale train/test/validation splits, ensuring comparability with prior methods. By introducing a dynamic, customizable benchmarking paradigm, Dynamic-KGQA enables a more rigorous and adaptable evaluation of QA systems.
☆ Biases in Large Language Model-Elicited Text: A Case Study in Natural Language Inference
We test whether NLP datasets created with Large Language Models (LLMs) contain annotation artifacts and social biases like NLP datasets elicited from crowd-source workers. We recreate a portion of the Stanford Natural Language Inference corpus using GPT-4, Llama-2 70b for Chat, and Mistral 7b Instruct. We train hypothesis-only classifiers to determine whether LLM-elicited NLI datasets contain annotation artifacts. Next, we use pointwise mutual information to identify the words in each dataset that are associated with gender, race, and age-related terms. On our LLM-generated NLI datasets, fine-tuned BERT hypothesis-only classifiers achieve between 86-96% accuracy. Our analyses further characterize the annotation artifacts and stereotypical biases in LLM-generated datasets.
comment: arXiv admin note: substantial text overlap with arXiv:2410.08996
☆ Provably Correct Automata Embeddings for Optimal Automata-Conditioned Reinforcement Learning
Automata-conditioned reinforcement learning (RL) has given promising results for learning multi-task policies capable of performing temporally extended objectives given at runtime, done by pretraining and freezing automata embeddings prior to training the downstream policy. However, no theoretical guarantees were given. This work provides a theoretical framework for the automata-conditioned RL problem and shows that it is probably approximately correct learnable. We then present a technique for learning provably correct automata embeddings, guaranteeing optimal multi-task policy learning. Our experimental evaluation confirms these theoretical results.
☆ Collapse of Dense Retrievers: Short, Early, and Literal Biases Outranking Factual Evidence
Dense retrieval models are commonly used in Information Retrieval (IR) applications, such as Retrieval-Augmented Generation (RAG). Since they often serve as the first step in these systems, their robustness is critical to avoid failures. In this work, by repurposing a relation extraction dataset (e.g. Re-DocRED), we design controlled experiments to quantify the impact of heuristic biases, such as favoring shorter documents, in retrievers like Dragon+ and Contriever. Our findings reveal significant vulnerabilities: retrievers often rely on superficial patterns like over-prioritizing document beginnings, shorter documents, repeated entities, and literal matches. Additionally, they tend to overlook whether the document contains the query's answer, lacking deep semantic understanding. Notably, when multiple biases combine, models exhibit catastrophic performance degradation, selecting the answer-containing document in less than 3% of cases over a biased document without the answer. Furthermore, we show that these biases have direct consequences for downstream applications like RAG, where retrieval-preferred documents can mislead LLMs, resulting in a 34% performance drop than not providing any documents at all.
☆ Continual Pre-training of MoEs: How robust is your router?
Sparsely-activated Mixture of Experts (MoE) transformers are promising architectures for foundation models. Compared to dense transformers that require the same amount of floating point operations (FLOPs) per forward pass, MoEs benefit from improved sample efficiency at training time and achieve much stronger performance. Many closed-source and open-source frontier language models have thus adopted an MoE architecture. Naturally, practitioners will want to extend the capabilities of these models with large amounts of newly collected data without completely re-training them. Prior work has shown that a simple combination of replay and learning rate re-warming and re-decaying can enable the continual pre-training (CPT) of dense decoder-only transformers with minimal performance degradation compared to full re-training. In the case of decoder-only MoE transformers, however, it is unclear how the routing algorithm will impact continual pre-training performance: 1) do the MoE transformer's routers exacerbate forgetting relative to a dense model?; 2) do the routers maintain a balanced load on previous distributions after CPT?; 3) are the same strategies applied to dense models sufficient to continually pre-train MoE LLMs? In what follows, we conduct a large-scale (>2B parameter switch and DeepSeek MoE LLMs trained for 600B tokens) empirical study across four MoE transformers to answer these questions. Our results establish a surprising robustness to distribution shifts for both Sinkhorn-Balanced and Z-and-Aux-loss-balanced routing algorithms, even in MoEs continually pre-trained without replay. Moreover, we show that MoE LLMs maintain their sample efficiency (relative to a FLOP-matched dense model) during CPT and that they can match the performance of a fully re-trained MoE at a fraction of the cost.
☆ Safety is Not Only About Refusal: Reasoning-Enhanced Fine-tuning for Interpretable LLM Safety
Large Language Models (LLMs) are vulnerable to jailbreak attacks that exploit weaknesses in traditional safety alignment, which often relies on rigid refusal heuristics or representation engineering to block harmful outputs. While they are effective for direct adversarial attacks, they fall short of broader safety challenges requiring nuanced, context-aware decision-making. To address this, we propose Reasoning-enhanced Finetuning for interpretable LLM Safety (Rational), a novel framework that trains models to engage in explicit safe reasoning before response. Fine-tuned models leverage the extensive pretraining knowledge in self-generated reasoning to bootstrap their own safety through structured reasoning, internalizing context-sensitive decision-making. Our findings suggest that safety extends beyond refusal, requiring context awareness for more robust, interpretable, and adaptive responses. Reasoning is not only a core capability of LLMs but also a fundamental mechanism for LLM safety. Rational employs reasoning-enhanced fine-tuning, allowing it to reject harmful prompts while providing meaningful and context-aware responses in complex scenarios.
☆ Leveraging Domain Knowledge at Inference Time for LLM Translation: Retrieval versus Generation
While large language models (LLMs) have been increasingly adopted for machine translation (MT), their performance for specialist domains such as medicine and law remains an open challenge. Prior work has shown that LLMs can be domain-adapted at test-time by retrieving targeted few-shot demonstrations or terminologies for inclusion in the prompt. Meanwhile, for general-purpose LLM MT, recent studies have found some success in generating similarly useful domain knowledge from an LLM itself, prior to translation. Our work studies domain-adapted MT with LLMs through a careful prompting setup, finding that demonstrations consistently outperform terminology, and retrieval consistently outperforms generation. We find that generating demonstrations with weaker models can close the gap with larger model's zero-shot performance. Given the effectiveness of demonstrations, we perform detailed analyses to understand their value. We find that domain-specificity is particularly important, and that the popular multi-domain benchmark is testing adaptation to a particular writing style more so than to a specific domain.
☆ Balcony: A Lightweight Approach to Dynamic Inference of Generative Language Models
Deploying large language models (LLMs) in real-world applications is often hindered by strict computational and latency constraints. While dynamic inference offers the flexibility to adjust model behavior based on varying resource budgets, existing methods are frequently limited by hardware inefficiencies or performance degradation. In this paper, we introduce Balcony, a simple yet highly effective framework for depth-based dynamic inference. By freezing the pretrained LLM and inserting additional transformer layers at selected exit points, Balcony maintains the full model's performance while enabling real-time adaptation to different computational budgets. These additional layers are trained using a straightforward self-distillation loss, aligning the sub-model outputs with those of the full model. This approach requires significantly fewer training tokens and tunable parameters, drastically reducing computational costs compared to prior methods. When applied to the LLaMA3-8B model, using only 0.2% of the original pretraining data, Balcony achieves minimal performance degradation while enabling significant speedups. Remarkably, we show that Balcony outperforms state-of-the-art methods such as Flextron and Layerskip as well as other leading compression techniques on multiple models and at various scales, across a variety of benchmarks.
☆ HieroLM: Egyptian Hieroglyph Recovery with Next Word Prediction Language Model NAACL 2025
Egyptian hieroglyphs are found on numerous ancient Egyptian artifacts, but it is common that they are blurry or even missing due to erosion. Existing efforts to restore blurry hieroglyphs adopt computer vision techniques such as CNNs and model hieroglyph recovery as an image classification task, which suffers from two major limitations: (i) They cannot handle severely damaged or completely missing hieroglyphs. (ii) They make predictions based on a single hieroglyph without considering contextual and grammatical information. This paper proposes a novel approach to model hieroglyph recovery as a next word prediction task and use language models to address it. We compare the performance of different SOTA language models and choose LSTM as the architecture of our HieroLM due to the strong local affinity of semantics in Egyptian hieroglyph texts. Experiments show that HieroLM achieves over 44% accuracy and maintains notable performance on multi-shot predictions and scarce data, which makes it a pragmatic tool to assist scholars in inferring missing hieroglyphs. It can also complement CV-based models to significantly reduce perplexity in recognizing blurry hieroglyphs. Our code is available at https://github.com/Rick-Cai/HieroLM/.
comment: Accepted at LaTeCH-CLfL 2025 @ NAACL 2025
☆ Wanda++: Pruning Large Language Models via Regional Gradients
Large Language Models (LLMs) pruning seeks to remove unimportant weights for inference speedup with minimal performance impact. However, existing methods often suffer from performance loss without full-model sparsity-aware fine-tuning. This paper presents Wanda++, a novel pruning framework that outperforms the state-of-the-art methods by utilizing decoder-block-level \textbf{regional} gradients. Specifically, Wanda++ improves the pruning score with regional gradients for the first time and proposes an efficient regional optimization method to minimize pruning-induced output discrepancies between the dense and sparse decoder output. Notably, Wanda++ improves perplexity by up to 32\% over Wanda in the language modeling task and generalizes effectively to downstream tasks. Further experiments indicate our proposed method is orthogonal to sparsity-aware fine-tuning, where Wanda++ can be combined with LoRA fine-tuning to achieve a similar perplexity improvement as the Wanda method. The proposed method is lightweight, pruning a 7B LLaMA model in under 10 minutes on a single NVIDIA H100 GPU.
☆ DP-GTR: Differentially Private Prompt Protection via Group Text Rewriting
Prompt privacy is crucial, especially when using online large language models (LLMs), due to the sensitive information often contained within prompts. While LLMs can enhance prompt privacy through text rewriting, existing methods primarily focus on document-level rewriting, neglecting the rich, multi-granular representations of text. This limitation restricts LLM utilization to specific tasks, overlooking their generalization and in-context learning capabilities, thus hindering practical application. To address this gap, we introduce DP-GTR, a novel three-stage framework that leverages local differential privacy (DP) and the composition theorem via group text rewriting. DP-GTR is the first framework to integrate both document-level and word-level information while exploiting in-context learning to simultaneously improve privacy and utility, effectively bridging local and global DP mechanisms at the individual data point level. Experiments on CommonSense QA and DocVQA demonstrate that DP-GTR outperforms existing approaches, achieving a superior privacy-utility trade-off. Furthermore, our framework is compatible with existing rewriting techniques, serving as a plug-in to enhance privacy protection. Our code is publicly available at https://github.com/FatShion-FTD/DP-GTR for reproducibility.
comment: 8 pages, 3 figures, 1 table
☆ Application of integrated gradients explainability to sociopsychological semantic markers
Classification of textual data in terms of sentiment, or more nuanced sociopsychological markers (e.g., agency), is now a popular approach commonly applied at the sentence level. In this paper, we exploit the integrated gradient (IG) method to capture the classification output at the word level, revealing which words actually contribute to the classification process. This approach improves explainability and provides in-depth insights into the text. We focus on sociopsychological markers beyond sentiment and investigate how to effectively train IG in agency, one of the very few markers for which a verified deep learning classifier, BERTAgent, is currently available. Performance and system parameters are carefully tested, alternatives to the IG approach are evaluated, and the usefulness of the result is verified in a relevant application scenario. The method is also applied in a scenario where only a small labeled dataset is available, with the aim of exploiting IG to identify the salient words that contribute to building the different classes that relate to relevant sociopsychological markers. To achieve this, an uncommon training procedure that encourages overfitting is employed to enhance the distinctiveness of each class. The results are analyzed through the lens of social psychology, offering valuable insights.
comment: Submitted to IEEE Trans. on Affective Computing
☆ LVLM-Compress-Bench: Benchmarking the Broader Impact of Large Vision-Language Model Compression NAACL 2025
Despite recent efforts in understanding the compression impact on large language models (LLMs) in terms of their downstream task performance and trustworthiness on relatively simpler uni-modal benchmarks (for example, question answering, common sense reasoning), their detailed study on multi-modal Large Vision-Language Models (LVLMs) is yet to be unveiled. Towards mitigating this gap, we present LVLM-Compress-Bench, a framework to first thoroughly study the broad impact of compression on the generative performance of LVLMs with multi-modal input driven tasks. In specific, we consider two major classes of compression for autoregressive models, namely KV cache and weight compression, for the dynamically growing intermediate cache and static weights, respectively. We use four LVLM variants of the popular LLaVA framework to present our analysis via integrating various state-of-the-art KV and weight compression methods including uniform, outlier-reduced, and group quantization for the KV cache and weights. With this framework we demonstrate on ten different multi-modal datasets with different capabilities including recognition, knowledge, language generation, spatial awareness, visual reasoning, hallucination and visual illusion identification, toxicity, stereotypes and bias. In specific, our framework demonstrates the compression impact on both general and ethically critical metrics leveraging a combination of real world and synthetic datasets to encompass diverse societal intersectional attributes. Extensive experimental evaluations yield diverse and intriguing observations on the behavior of LVLMs at different quantization budget of KV and weights, in both maintaining and losing performance as compared to the baseline model with FP16 data format. Code will be open-sourced at https://github.com/opengear-project/LVLM-compress-bench.
comment: This work has been accepted to NAACL 2025
☆ Beyond RAG: Task-Aware KV Cache Compression for Comprehensive Knowledge Reasoning
Incorporating external knowledge in large language models (LLMs) enhances their utility across diverse applications, but existing methods have trade-offs. Retrieval-Augmented Generation (RAG) fetches evidence via similarity search, but key information may fall outside top ranked results. Long-context models can process multiple documents but are computationally expensive and limited by context window size. Inspired by students condensing study material for open-book exams, we propose task-aware key-value (KV) cache compression, which compresses external knowledge in a zero- or few-shot setup. This enables LLMs to reason efficiently over a compacted representation of all relevant information. Experiments show our approach outperforms both RAG and task-agnostic compression methods. On LongBench v2, it improves accuracy by up to 7 absolute points over RAG with a 30x compression rate, while reducing inference latency from 0.43s to 0.16s. A synthetic dataset highlights that RAG performs well when sparse evidence suffices, whereas task-aware compression is superior for broad knowledge tasks.
☆ Evaluating Answer Reranking Strategies in Time-sensitive Question Answering
Despite advancements in state-of-the-art models and information retrieval techniques, current systems still struggle to handle temporal information and to correctly answer detailed questions about past events. In this paper, we investigate the impact of temporal characteristics of answers in Question Answering (QA) by exploring several simple answer selection techniques. Our findings emphasize the role of temporal features in selecting the most relevant answers from diachronic document collections and highlight differences between explicit and implicit temporal questions.
☆ DB-Explore: Automated Database Exploration and Instruction Synthesis for Text-to-SQL
Recent text-to-SQL systems powered by large language models (LLMs) have demonstrated remarkable performance in translating natural language queries into SQL. However, these systems often struggle with complex database structures and domain-specific queries, as they primarily focus on enhancing logical reasoning and SQL syntax while overlooking the critical need for comprehensive database understanding. To address this limitation, we propose DB-Explore, a novel framework that systematically aligns LLMs with database knowledge through automated exploration and instruction synthesis. DB-Explore constructs database graphs to capture complex relational schemas, leverages GPT-4 to systematically mine structural patterns and semantic knowledge, and synthesizes instructions to distill this knowledge for efficient fine-tuning of LLMs. Our framework enables comprehensive database understanding through diverse sampling strategies and automated instruction generation, bridging the gap between database structures and language models. Experiments conducted on the SPIDER and BIRD benchmarks validate the effectiveness of DB-Explore, achieving an execution accuracy of 52.1% on BIRD and 84.0% on SPIDER. Notably, our open-source implementation, based on the Qwen2.5-coder-7B model, outperforms multiple GPT-4-driven text-to-SQL systems in comparative evaluations, and achieves near state-of-the-art performance with minimal computational cost.
☆ SafeArena: Evaluating the Safety of Autonomous Web Agents
LLM-based agents are becoming increasingly proficient at solving web-based tasks. With this capability comes a greater risk of misuse for malicious purposes, such as posting misinformation in an online forum or selling illicit substances on a website. To evaluate these risks, we propose SafeArena, the first benchmark to focus on the deliberate misuse of web agents. SafeArena comprises 250 safe and 250 harmful tasks across four websites. We classify the harmful tasks into five harm categories -- misinformation, illegal activity, harassment, cybercrime, and social bias, designed to assess realistic misuses of web agents. We evaluate leading LLM-based web agents, including GPT-4o, Claude-3.5 Sonnet, Qwen-2-VL 72B, and Llama-3.2 90B, on our benchmark. To systematically assess their susceptibility to harmful tasks, we introduce the Agent Risk Assessment framework that categorizes agent behavior across four risk levels. We find agents are surprisingly compliant with malicious requests, with GPT-4o and Qwen-2 completing 34.7% and 27.3% of harmful requests, respectively. Our findings highlight the urgent need for safety alignment procedures for web agents. Our benchmark is available here: https://safearena.github.io
☆ Collaborative Evaluation of Deepfake Text with Deliberation-Enhancing Dialogue Systems
The proliferation of generative models has presented significant challenges in distinguishing authentic human-authored content from deepfake content. Collaborative human efforts, augmented by AI tools, present a promising solution. In this study, we explore the potential of DeepFakeDeLiBot, a deliberation-enhancing chatbot, to support groups in detecting deepfake text. Our findings reveal that group-based problem-solving significantly improves the accuracy of identifying machine-generated paragraphs compared to individual efforts. While engagement with DeepFakeDeLiBot does not yield substantial performance gains overall, it enhances group dynamics by fostering greater participant engagement, consensus building, and the frequency and diversity of reasoning-based utterances. Additionally, participants with higher perceived effectiveness of group collaboration exhibited performance benefits from DeepFakeDeLiBot. These findings underscore the potential of deliberative chatbots in fostering interactive and productive group dynamics while ensuring accuracy in collaborative deepfake text detection. \textit{Dataset and source code used in this study will be made publicly available upon acceptance of the manuscript.
comment: 15
☆ VQEL: Enabling Self-Developed Symbolic Language in Agents through Vector Quantization in Emergent Language Games
In the field of emergent language, efforts have traditionally focused on developing communication protocols through interactions between agents in referential games. However, the aspect of internal language learning, where language serves not only as a communicative tool with others but also as a means for individual thinking, self-reflection, and problem-solving remains underexplored. Developing a language through self-play, without another agent's involvement, poses a unique challenge. It requires an agent to craft symbolic representations and train them using direct gradient methods. The challenge here is that if an agent attempts to learn symbolic representations through self-play using conventional modeling and techniques such as REINFORCE, the solution will offer no advantage over previous multi-agent approaches. We introduce VQEL, a novel method that incorporates Vector Quantization into the agents' architecture, enabling them to autonomously invent and develop discrete symbolic representations in a self-play referential game. Following the self-play phase, agents can enhance their language through reinforcement learning and interactions with other agents in the mutual-play phase. Our experiments across various datasets demonstrate that VQEL not only outperforms the traditional REINFORCE method but also benefits from improved control and reduced susceptibility to collapse, thanks to the incorporation of vector quantization.
☆ HILGEN: Hierarchically-Informed Data Generation for Biomedical NER Using Knowledgebases and Large Language Models
We present HILGEN, a Hierarchically-Informed Data Generation approach that combines domain knowledge from the Unified Medical Language System (UMLS) with synthetic data generated by large language models (LLMs), specifically GPT-3.5. Our approach leverages UMLS's hierarchical structure to expand training data with related concepts, while incorporating contextual information from LLMs through targeted prompts aimed at automatically generating synthetic examples for sparsely occurring named entities. The performance of the HILGEN approach was evaluated across four biomedical NER datasets (MIMIC III, BC5CDR, NCBI-Disease, and Med-Mentions) using BERT-Large and DANN (Data Augmentation with Nearest Neighbor Classifier) models, applying various data generation strategies, including UMLS, GPT-3.5, and their best ensemble. For the BERT-Large model, incorporating UMLS led to an average F1 score improvement of 40.36%, while using GPT-3.5 resulted in a comparable average increase of 40.52%. The Best-Ensemble approach using BERT-Large achieved the highest improvement, with an average increase of 42.29%. DANN model's F1 score improved by 22.74% on average using the UMLS-only approach. The GPT-3.5-based method resulted in a 21.53% increase, and the Best-Ensemble DANN model showed a more notable improvement, with an average increase of 25.03%. Our proposed HILGEN approach improves NER performance in few-shot settings without requiring additional manually annotated data. Our experiments demonstrate that an effective strategy for optimizing biomedical NER is to combine biomedical knowledge curated in the past, such as the UMLS, and generative LLMs to create synthetic training instances. Our future research will focus on exploring additional innovative synthetic data generation strategies for further improving NER performance.
☆ Maximizing Signal in Human-Model Preference Alignment AAAI 2025
The emergence of powerful LLMs has led to a paradigm shift in Natural Language Understanding and Natural Language Generation. The properties that make LLMs so valuable for these tasks -- creativity, ability to produce fluent speech, and ability to quickly and effectively abstract information from large corpora -- also present new challenges to evaluating their outputs. The rush to market has led teams to fall back on quick, cost-effective automatic evaluations which offer value, but do not obviate the need for human judgments in model training and evaluation. This paper argues that in cases in which end users need to agree with the decisions made by ML models -- e.g. in toxicity detection or extraction of main points for summarization -- models should be trained and evaluated on data that represent the preferences of those users. We support this argument by explicating the role of human feedback in labeling and judgment tasks for model training and evaluation. First, we propose methods for disentangling noise from signal in labeling tasks. Then we show that noise in labeling disagreement can be minimized by adhering to proven methodological best practices, while signal can be maximized to play an integral role in model training and evaluation tasks. Finally, we illustrate best practices by providing a case study in which two guardrails classifiers are evaluated using human judgments to align final model behavior to user preferences. We aim for this paper to provide researchers and professionals with guidelines to integrating human judgments into their ML and generative AI evaluation toolkit, particularly when working toward achieving accurate and unbiased features that align with users' needs and expectations.
comment: Presented at AAAI 2025, special track on AI Alignment
♻ ☆ Unified Mind Model: Reimagining Autonomous Agents in the LLM Era
Large language models (LLMs) have recently demonstrated remarkable capabilities across domains, tasks, and languages (e.g., ChatGPT and GPT-4), reviving the research of general autonomous agents with human-like cognitive abilities. Such human-level agents require semantic comprehension and instruction-following capabilities, which exactly fall into the strengths of LLMs. Although there have been several initial attempts to build human-level agents based on LLMs, the theoretical foundation remains a challenging open problem. In this paper, we propose a novel theoretical cognitive architecture, the Unified Mind Model (UMM), which offers guidance to facilitate the rapid creation of autonomous agents with human-level cognitive abilities. Specifically, our UMM starts with the global workspace theory and further leverage LLMs to enable the agent with various cognitive abilities, such as multi-modal perception, planning, reasoning, tool use, learning, memory, reflection and motivation. Building upon UMM, we then develop an agent-building engine, MindOS, which allows users to quickly create domain-/task-specific autonomous agents without any programming effort.
comment: 18 pages
♻ ☆ Explaining Caption-Image Interactions in CLIP models with Second-Order Attributions
Dual encoder architectures like CLIP models map two types of inputs into a shared embedding space and predict similarities between them. Despite their success, it is, however, not understood how these models compare their two inputs. Common first-order feature-attribution methods can only provide limited insights into dual-encoders since their predictions depend on feature-interactions rather than on individual features. In this paper, we first derive a second-order method enabling the attribution of predictions by any differentiable dual encoder onto feature-interactions between its inputs. Second, we apply our method to CLIP models and show that they learn fine-grained correspondences between parts of captions and regions in images. They match objects across input modes also account for mismatches. This visual-linguistic grounding ability, however, varies heavily between object classes and exhibits pronounced out-of-domain effects. We can identify individual errors as well as systematic failure categories including object coverage, unusual scenes and correlated contexts.
♻ ☆ Gated Delta Networks: Improving Mamba2 with Delta Rule ICLR 2025
Linear Transformers have gained attention as efficient alternatives to standard Transformers, but their performance in retrieval and long-context tasks has been limited. To address these limitations, recent work has explored two distinct mechanisms: gating for adaptive memory control and the delta update rule for precise memory modifications. We observe that these mechanisms are complementary: gating enables rapid memory erasure while the delta rule facilitates targeted updates. Building on this insight, we introduce the gated delta rule and develop a parallel training algorithm optimized for modern hardware. Our proposed architecture, Gated DeltaNet, consistently surpasses existing models like Mamba2 and DeltaNet across multiple benchmarks, including language modeling, common-sense reasoning, in-context retrieval, length extrapolation, and long-context understanding. We further enhance performance by developing hybrid architectures that combine Gated DeltaNet layers with sliding window attention or Mamba2 layers, achieving both improved training efficiency and superior task performance.
comment: ICLR 2025 camera ready
♻ ☆ How Far Are We on the Decision-Making of LLMs? Evaluating LLMs' Gaming Ability in Multi-Agent Environments ICLR 2025
Decision-making is a complex process requiring diverse abilities, making it an excellent framework for evaluating Large Language Models (LLMs). Researchers have examined LLMs' decision-making through the lens of Game Theory. However, existing evaluation mainly focus on two-player scenarios where an LLM competes against another. Additionally, previous benchmarks suffer from test set leakage due to their static design. We introduce GAMA($\gamma$)-Bench, a new framework for evaluating LLMs' Gaming Ability in Multi-Agent environments. It includes eight classical game theory scenarios and a dynamic scoring scheme specially designed to quantitatively assess LLMs' performance. $\gamma$-Bench allows flexible game settings and adapts the scoring system to different game parameters, enabling comprehensive evaluation of robustness, generalizability, and strategies for improvement. Our results indicate that GPT-3.5 demonstrates strong robustness but limited generalizability, which can be enhanced using methods like Chain-of-Thought. We also evaluate 13 LLMs from 6 model families, including GPT-3.5, GPT-4, Gemini, LLaMA-3.1, Mixtral, and Qwen-2. Gemini-1.5-Pro outperforms others, scoring of $69.8$ out of $100$, followed by LLaMA-3.1-70B ($65.9$) and Mixtral-8x22B ($62.4$). Our code and experimental results are publicly available at https://github.com/CUHK-ARISE/GAMABench.
comment: Accepted to ICLR 2025; 11 pages of main text; 26 pages of appendices; Included models: GPT-3.5-{0613, 1106, 0125}, GPT-4-0125, GPT-4o-0806, Gemini-{1.0, 1.5)-Pro, LLaMA-3.1-{7, 70, 405}B, Mixtral-8x{7, 22}B, Qwen-2-72B
♻ ☆ HELMET: How to Evaluate Long-Context Language Models Effectively and Thoroughly ICLR 2025
Many benchmarks exist for evaluating long-context language models (LCLMs), yet developers often rely on synthetic tasks such as needle-in-a-haystack (NIAH) or an arbitrary subset of tasks. However, it remains unclear whether these benchmarks reflect the diverse downstream applications of LCLMs, and such inconsistencies further complicate model comparison. We investigate the underlying reasons behind these practices and find that existing benchmarks often provide noisy signals due to limited coverage of applications, insufficient context lengths, unreliable metrics, and incompatibility with base models. In this work, we introduce HELMET (How to Evaluate Long-context Models Effectively and Thoroughly), a comprehensive benchmark encompassing seven diverse, application-centric categories. We also address several issues in previous benchmarks by adding controllable lengths up to 128K tokens, model-based evaluation for reliable metrics, and few-shot prompting for robustly evaluating base models. Consequently, we demonstrate that HELMET offers more reliable and consistent rankings of frontier LCLMs. Through a comprehensive study of 59 LCLMs, we find that (1) synthetic tasks like NIAH do not reliably predict downstream performance; (2) the diverse categories in HELMET exhibit distinct trends and low correlations with each other; and (3) while most LCLMs achieve perfect NIAH scores, open-source models significantly lag behind closed ones when tasks require full-context reasoning or following complex instructions -- the gap widens as length increases. Finally, we recommend using our RAG tasks for fast model development, as they are easy to run and better predict other downstream performance; ultimately, we advocate for a holistic evaluation across diverse tasks.
comment: ICLR 2025. Project page: https://princeton-nlp.github.io/HELMET/
♻ ☆ AdaptBot: Combining LLM with Knowledge Graphs and Human Input for Generic-to-Specific Task Decomposition and Knowledge Refinement ICRA
An embodied agent assisting humans is often asked to complete new tasks, and there may not be sufficient time or labeled examples to train the agent to perform these new tasks. Large Language Models (LLMs) trained on considerable knowledge across many domains can be used to predict a sequence of abstract actions for completing such tasks, although the agent may not be able to execute this sequence due to task-, agent-, or domain-specific constraints. Our framework addresses these challenges by leveraging the generic predictions provided by LLM and the prior domain knowledge encoded in a Knowledge Graph (KG), enabling an agent to quickly adapt to new tasks. The robot also solicits and uses human input as needed to refine its existing knowledge. Based on experimental evaluation in the context of cooking and cleaning tasks in simulation domains, we demonstrate that the interplay between LLM, KG, and human input leads to substantial performance gains compared with just using the LLM. Project website{\S}: https://sssshivvvv.github.io/adaptbot/
comment: Accepted to IEEE International Conference on Robotics and Automation (ICRA) 2025
♻ ☆ Diagnosing Moral Reasoning Acquisition in Language Models: Pragmatics and Generalization
Ensuring that Large Language Models (LLMs) return just responses which adhere to societal values is crucial for their broader application. Prior research has shown that LLMs often fail to perform satisfactorily on tasks requiring moral cognizance, such as ethics-based judgments. While current approaches have focused on fine-tuning LLMs with curated datasets to improve their capabilities on such tasks, choosing the optimal learning paradigm to enhance the ethical responses of LLMs remains an open research debate. In this work, we aim to address this fundamental question: can current learning paradigms enable LLMs to acquire sufficient moral reasoning capabilities? Drawing from distributional semantics theory and the pragmatic nature of moral discourse, our analysis indicates that performance improvements follow a mechanism similar to that of semantic-level tasks, and therefore remain affected by the pragmatic nature of morals latent in discourse, a phenomenon we name the pragmatic dilemma. We conclude that this pragmatic dilemma imposes significant limitations on the generalization ability of current learning paradigms, making it the primary bottleneck for moral reasoning acquisition in LLMs.
♻ ☆ Get my drift? Catching LLM Task Drift with Activation Deltas
LLMs are commonly used in retrieval-augmented applications to execute user instructions based on data from external sources. For example, modern search engines use LLMs to answer queries based on relevant search results; email plugins summarize emails by processing their content through an LLM. However, the potentially untrusted provenance of these data sources can lead to prompt injection attacks, where the LLM is manipulated by natural language instructions embedded in the external data, causing it to deviate from the user's original instruction(s). We define this deviation as task drift. Task drift is a significant concern as it allows attackers to exfiltrate data or influence the LLM's output for other users. We study LLM activations as a solution to detect task drift, showing that activation deltas - the difference in activations before and after processing external data - are strongly correlated with this phenomenon. Through two probing methods, we demonstrate that a simple linear classifier can detect drift with near-perfect ROC AUC on an out-of-distribution test set. We evaluate these methods by making minimal assumptions about how users' tasks, system prompts, and attacks can be phrased. We observe that this approach generalizes surprisingly well to unseen task domains, such as prompt injections, jailbreaks, and malicious instructions, without being trained on any of these attacks. Interestingly, the fact that this solution does not require any modifications to the LLM (e.g., fine-tuning), as well as its compatibility with existing meta-prompting solutions, makes it cost-efficient and easy to deploy. To encourage further research on activation-based task inspection, decoding, and interpretability, we release our large-scale TaskTracker toolkit, featuring a dataset of over 500K instances, representations from six SoTA language models, and a suite of inspection tools.
comment: SaTML 2025
♻ ☆ ACC-Collab: An Actor-Critic Approach to Multi-Agent LLM Collaboration
Large language models (LLMs) have demonstrated a remarkable ability to serve as general-purpose tools for various language-based tasks. Recent works have demonstrated that the efficacy of such models can be improved through iterative dialog between multiple models. While these paradigms show promise in improving model efficacy, most works in this area treat collaboration as an emergent behavior, rather than a learned behavior. In doing so, current multi-agent frameworks rely on collaborative behaviors to have been sufficiently trained into off-the-shelf models. To address this limitation, we propose ACC-Collab, an Actor-Critic based learning framework to produce a two-agent team (an actor-agent and a critic-agent) specialized in collaboration. We demonstrate that ACC-Collab outperforms SotA multi-agent techniques on a wide array of benchmarks.
♻ ☆ LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation
Assessing the reasoning capabilities of large language models (LLMs) is susceptible to overestimation due to data exposure of evaluation benchmarks. We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates and apply this framework to develop LINGOLY-TOO, a challenging benchmark for linguistic reasoning. By developing orthographic templates, we dynamically obfuscate the writing systems of real languages to generate numerousquestion variations. These variations preserve the reasoning steps required for each solution while reducing the likelihood of specific problem instances appearing in model training data. Our experiments demonstrate that frontier models, including Claud 3.7 Sonnet, o1-preview and DeepSeek R1, struggle with advanced reasoning. Our analysis also shows that LLMs exhibit noticeable variance in accuracy across permutations of the same problem, and on average perform better on questions appearing in their original orthography. Our findings highlight the opaque nature of response generation in LLMs and provide evidence that prior data exposure contributes to over estimating the reasoning capabilities of frontier models.
♻ ☆ Protein Large Language Models: A Comprehensive Survey
Protein-specific large language models (Protein LLMs) are revolutionizing protein science by enabling more efficient protein structure prediction, function annotation, and design. While existing surveys focus on specific aspects or applications, this work provides the first comprehensive overview of Protein LLMs, covering their architectures, training datasets, evaluation metrics, and diverse applications. Through a systematic analysis of over 100 articles, we propose a structured taxonomy of state-of-the-art Protein LLMs, analyze how they leverage large-scale protein sequence data for improved accuracy, and explore their potential in advancing protein engineering and biomedical research. Additionally, we discuss key challenges and future directions, positioning Protein LLMs as essential tools for scientific discovery in protein science. Resources are maintained at https://github.com/Yijia-Xiao/Protein-LLM-Survey.
comment: 24 pages, 4 figures, 5 tables
♻ ☆ NormAd: A Framework for Measuring the Cultural Adaptability of Large Language Models NAACL 2025
To be effectively and safely deployed to global user populations, large language models (LLMs) may need to adapt outputs to user values and cultures, not just know about them. We introduce NormAd, an evaluation framework to assess LLMs' cultural adaptability, specifically measuring their ability to judge social acceptability across varying levels of cultural norm specificity, from abstract values to explicit social norms. As an instantiation of our framework, we create NormAd-Eti, a benchmark of 2.6k situational descriptions representing social-etiquette related cultural norms from 75 countries. Through comprehensive experiments on NormAd-Eti, we find that LLMs struggle to accurately judge social acceptability across these varying degrees of cultural contexts and show stronger adaptability to English-centric cultures over those from the Global South. Even in the simplest setting where the relevant social norms are provided, the best LLMs' performance (< 82\%) lags behind humans (> 95\%). In settings with abstract values and country information, model performance drops substantially (< 60\%), while human accuracy remains high (> 90\%). Furthermore, we find that models are better at recognizing socially acceptable versus unacceptable situations. Our findings showcase the current pitfalls in socio-cultural reasoning of LLMs which hinder their adaptability for global audiences.
comment: Accepted at NAACL 2025
♻ ☆ $\texttt{SEM-CTRL}$: Semantically Controlled Decoding
Ensuring both syntactic and semantic correctness in Large Language Model (LLM) outputs remains a significant challenge, despite being critical for real-world deployment. In this paper, we introduce $\texttt{SEM-CTRL}$, a unified approach that enforces rich context-sensitive constraints and task- and instance-specific semantics directly on an LLM decoder. Our approach integrates token-level MCTS, which is guided by specific syntactic and semantic constraints. The constraints over the desired outputs are expressed using Answer Set Grammars -- a logic-based formalism that generalizes context-sensitive grammars while incorporating background knowledge to represent task-specific semantics. We show that our approach guarantees correct completions for any off-the-shelf LLM without the need for fine-tuning. We evaluate $\texttt{SEM-CTRL}$ on a range of tasks, including synthetic grammar synthesis, combinatorial reasoning, and planning. Our results demonstrate that $\texttt{SEM-CTRL}$ allows small pre-trained LLMs to efficiently outperform larger variants and state-of-the-art reasoning models (e.g., o1-preview) while simultaneously guaranteeing solution correctness.
♻ ☆ Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution ICLR 2025
Probing learned concepts in large language models (LLMs) is crucial for understanding how semantic knowledge is encoded internally. Training linear classifiers on probing tasks is a principle approach to denote the vector of a certain concept in the representation space. However, the single vector identified for a concept varies with both data and training, making it less robust and weakening its effectiveness in real-world applications. To address this challenge, we propose an approach to approximate the subspace representing a specific concept. Built on linear probing classifiers, we extend the concept vectors into Gaussian Concept Subspace (GCS). We demonstrate GCS's effectiveness through measuring its faithfulness and plausibility across multiple LLMs with different sizes and architectures. Additionally, we use representation intervention tasks to showcase its efficacy in real-world applications such as emotion steering. Experimental results indicate that GCS concept vectors have the potential to balance steering performance and maintaining the fluency in natural language generation tasks.
comment: Accepted by ICLR 2025
♻ ☆ X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability
Despite the rapid development of safety alignment techniques for LLMs, defending against multi-turn jailbreaks is still a challenging task. In this paper, we conduct a comprehensive comparison, revealing that some existing defense methods can improve the robustness of LLMs against multi-turn jailbreaks but compromise usability, i.e., reducing general capabilities or causing the over-refusal problem. From the perspective of mechanism interpretability of LLMs, we discover that these methods fail to establish a boundary that exactly distinguishes safe and harmful feature representations. Therefore, boundary-safe representations close to harmful representations are inevitably disrupted, leading to a decline in usability. To address this issue, we propose X-Boundary to push harmful representations away from boundary-safe representations and obtain an exact distinction boundary. In this way, harmful representations can be precisely erased without disrupting safe ones. Experimental results show that X-Boundary achieves state-of-the-art defense performance against multi-turn jailbreaks, while reducing the over-refusal rate by about 20% and maintaining nearly complete general capability. Furthermore, we theoretically prove and empirically verify that X-Boundary can accelerate the convergence process during training. Please see our code at: https://github.com/AI45Lab/X-Boundary.
♻ ☆ UoR-NCL at SemEval-2025 Task 1: Using Generative LLMs and CLIP Models for Multilingual Multimodal Idiomaticity Representation
SemEval-2025 Task 1 focuses on ranking images based on their alignment with a given nominal compound that may carry idiomatic meaning in both English and Brazilian Portuguese. To address this challenge, this work uses generative large language models (LLMs) and multilingual CLIP models to enhance idiomatic compound representations. LLMs generate idiomatic meanings for potentially idiomatic compounds, enriching their semantic interpretation. These meanings are then encoded using multilingual CLIP models, serving as representations for image ranking. Contrastive learning and data augmentation techniques are applied to fine-tune these embeddings for improved performance. Experimental results show that multimodal representations extracted through this method outperformed those based solely on the original nominal compounds. The fine-tuning approach shows promising outcomes but is less effective than using embeddings without fine-tuning. The source code used in this paper is available at https://github.com/tongwu17/SemEval-2025-Task1-UoR-NCL.
♻ ☆ Gumbel Counterfactual Generation From Language Models ICLR 2025
Understanding and manipulating the causal generation mechanisms in language models is essential for controlling their behavior. Previous work has primarily relied on techniques such as representation surgery -- e.g., model ablations or manipulation of linear subspaces tied to specific concepts -- to \emph{intervene} on these models. To understand the impact of interventions precisely, it is useful to examine \emph{counterfactuals} -- e.g., how a given sentence would have appeared had it been generated by the model following a specific intervention. We highlight that counterfactual reasoning is conceptually distinct from interventions, as articulated in Pearl's causal hierarchy. Based on this observation, we propose a framework for generating true string counterfactuals by reformulating language models as a structural equation model using the Gumbel-max trick, which we called Gumbel counterfactual generation. This reformulation allows us to model the joint distribution over original strings and their counterfactuals resulting from the same instantiation of the sampling noise. We develop an algorithm based on hindsight Gumbel sampling that allows us to infer the latent noise variables and generate counterfactuals of observed strings. Our experiments demonstrate that the approach produces meaningful counterfactuals while at the same time showing that commonly used intervention techniques have considerable undesired side effects.
comment: Accepted in ICLR 2025
♻ ☆ Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models ICLR 2025
The capabilities and limitations of Large Language Models have been sketched out in great detail in recent years, providing an intriguing yet conflicting picture. On the one hand, LLMs demonstrate a general ability to solve problems. On the other hand, they show surprising reasoning gaps when compared to humans, casting doubt on the robustness of their generalisation strategies. The sheer volume of data used in the design of LLMs has precluded us from applying the method traditionally used to measure generalisation: train-test set separation. To overcome this, we study what kind of generalisation strategies LLMs employ when performing reasoning tasks by investigating the pretraining data they rely on. For two models of different sizes (7B and 35B) and 2.5B of their pretraining tokens, we identify what documents influence the model outputs for three simple mathematical reasoning tasks and contrast this to the data that are influential for answering factual questions. We find that, while the models rely on mostly distinct sets of data for each factual question, a document often has a similar influence across different reasoning questions within the same task, indicating the presence of procedural knowledge. We further find that the answers to factual questions often show up in the most influential data. However, for reasoning questions the answers usually do not show up as highly influential, nor do the answers to the intermediate reasoning steps. When we characterise the top ranked documents for the reasoning questions qualitatively, we confirm that the influential documents often contain procedural knowledge, like demonstrating how to obtain a solution using formulae or code. Our findings indicate that the approach to reasoning the models use is unlike retrieval, and more like a generalisable strategy that synthesises procedural knowledge from documents doing a similar form of reasoning.
comment: Published at ICLR 2025
♻ ☆ Approaching the Limits to EFL Writing Enhancement with AI-generated Text and Diverse Learners
Generative artificial intelligence (AI) chatbots, such as ChatGPT, are reshaping how English as a foreign language (EFL) students write since students can compose texts by integrating their own words with AI-generated text. This study investigated how 59 Hong Kong secondary school students with varying levels of academic achievement interacted with AI-generated text to compose a feature article, exploring whether any interaction patterns benefited the overall quality of the article. Through content analysis, multiple linear regression and cluster analysis, we found the overall number of words -- whether AI- or human-generated -- is the main predictor of writing quality. However, the impact varies by students' competence to write independently, for instance, by using their own words accurately and coherently to compose a text, and to follow specific interaction patterns with AI-generated text. Therefore, although composing texts with human words and AI-generated text may become prevalent in EFL writing classrooms, without educators' careful attention to EFL writing pedagogy and AI literacy, high-achieving students stand to benefit more from using AI-generated text than low-achieving students.
♻ ☆ Assisting Mathematical Formalization with A Learning-based Premise Retriever
Premise selection is a crucial yet challenging step in mathematical formalization, especially for users with limited experience. Due to the lack of available formalization projects, existing approaches that leverage language models often suffer from data scarcity. In this work, we introduce an innovative method for training a premise retriever to support the formalization of mathematics. Our approach employs a BERT model to embed proof states and premises into a shared latent space. The retrieval model is trained within a contrastive learning framework and incorporates a domain-specific tokenizer along with a fine-grained similarity computation method. Experimental results show that our model is highly competitive compared to existing baselines, achieving strong performance while requiring fewer computational resources. Performance is further enhanced through the integration of a re-ranking module. To streamline the formalization process, we will release a search engine that enables users to query Mathlib theorems directly using proof states, significantly improving accessibility and efficiency. Codes are available at https://github.com/ruc-ai4math/Premise-Retrieval.
♻ ☆ AfroBench: How Good are Large Language Models on African Languages?
Large-scale multilingual evaluations, such as MEGA, often include only a handful of African languages due to the scarcity of high-quality evaluation data and the limited discoverability of existing African datasets. This lack of representation hinders comprehensive LLM evaluation across a diverse range of languages and tasks. To address these challenges, we introduce AfroBench -- a multi-task benchmark for evaluating the performance of LLMs across 64 African languages, 15 tasks and 22 datasets. AfroBench consists of nine natural language understanding datasets, six text generation datasets, six knowledge and question answering tasks, and one mathematical reasoning task. We present results comparing the performance of prompting LLMs to fine-tuned baselines based on BERT and T5-style models. Our results suggest large gaps in performance between high-resource languages, such as English, and African languages across most tasks; but performance also varies based on the availability of monolingual data resources. Our findings confirm that performance on African languages continues to remain a hurdle for current LLMs, underscoring the need for additional efforts to close this gap. https://mcgill-nlp.github.io/AfroBench/
comment: Under review
♻ ☆ OlympicArena: Benchmarking Multi-discipline Cognitive Reasoning for Superintelligent AI NeurIPS 2024
The evolution of Artificial Intelligence (AI) has been significantly accelerated by advancements in Large Language Models (LLMs) and Large Multimodal Models (LMMs), gradually showcasing potential cognitive reasoning abilities in problem-solving and scientific discovery (i.e., AI4Science) once exclusive to human intellect. To comprehensively evaluate current models' performance in cognitive reasoning abilities, we introduce OlympicArena, which includes 11,163 bilingual problems across both text-only and interleaved text-image modalities. These challenges encompass a wide range of disciplines spanning seven fields and 62 international Olympic competitions, rigorously examined for data leakage. We argue that the challenges in Olympic competition problems are ideal for evaluating AI's cognitive reasoning due to their complexity and interdisciplinary nature, which are essential for tackling complex scientific challenges and facilitating discoveries. Beyond evaluating performance across various disciplines using answer-only criteria, we conduct detailed experiments and analyses from multiple perspectives. We delve into the models' cognitive reasoning abilities, their performance across different modalities, and their outcomes in process-level evaluations, which are vital for tasks requiring complex reasoning with lengthy solutions. Our extensive evaluations reveal that even advanced models like GPT-4o only achieve a 39.97% overall accuracy, illustrating current AI limitations in complex reasoning and multimodal integration. Through the OlympicArena, we aim to advance AI towards superintelligence, equipping it to address more complex challenges in science and beyond. We also provide a comprehensive set of resources to support AI research, including a benchmark dataset, an open-source annotation platform, a detailed evaluation tool, and a leaderboard with automatic submission features.
comment: Accepted by NeurIPS 2024
♻ ☆ 360$^\circ$REA: Towards A Reusable Experience Accumulation with 360° Assessment for Multi-Agent System
Large language model agents have demonstrated remarkable advancements across various complex tasks. Recent works focus on optimizing the agent team or employing self-reflection to iteratively solve complex tasks. Since these agents are all based on the same LLM, only conducting self-evaluation or removing underperforming agents does not substantively enhance the capability of the agents. We argue that a comprehensive evaluation and accumulating experience from evaluation feedback is an effective approach to improving system performance. In this paper, we propose Reusable Experience Accumulation with 360$^\circ$ Assessment (360$^\circ$REA), a hierarchical multi-agent framework inspired by corporate organizational practices. The framework employs a novel 360$^\circ$ performance assessment method for multi-perspective performance evaluation with fine-grained assessment. To enhance the capability of agents in addressing complex tasks, we introduce dual-level experience pool for agents to accumulate experience through fine-grained assessment. Extensive experiments on complex task datasets demonstrate the effectiveness of 360$^\circ$REA.
♻ ☆ Structured Preference Optimization for Vision-Language Long-Horizon Task Planning
Existing methods for vision-language task planning excel in short-horizon tasks but often fall short in complex, long-horizon planning within dynamic environments. These challenges primarily arise from the difficulty of effectively training models to produce high-quality reasoning processes for long-horizon tasks. To address this, we propose Structured Preference Optimization (SPO), which aims to enhance reasoning and action selection in long-horizon task planning through structured preference evaluation and optimized training strategies. Specifically, SPO introduces: 1) Preference-Based Scoring and Optimization, which systematically evaluates reasoning chains based on task relevance, visual grounding, and historical consistency; and 2) Curriculum-Guided Training, where the model progressively adapts from simple to complex tasks, improving its generalization ability in long-horizon scenarios and enhancing reasoning robustness. To advance research in vision-language long-horizon task planning, we introduce ExtendaBench, a comprehensive benchmark covering 1,509 tasks across VirtualHome and Habitat 2.0, categorized into ultra-short, short, medium, and long tasks. Experimental results demonstrate that SPO significantly improves reasoning quality and final decision accuracy, outperforming prior methods on long-horizon tasks and underscoring the effectiveness of preference-driven optimization in vision-language task planning. Specifically, SPO achieves a +5.98% GCR and +4.68% SR improvement in VirtualHome and a +3.30% GCR and +2.11% SR improvement in Habitat over the best-performing baselines.
comment: 18 pages
♻ ☆ Stealthy Jailbreak Attacks on Large Language Models via Benign Data Mirroring NAACL 2025
Large language model (LLM) safety is a critical issue, with numerous studies employing red team testing to enhance model security. Among these, jailbreak methods explore potential vulnerabilities by crafting malicious prompts that induce model outputs contrary to safety alignments. Existing black-box jailbreak methods often rely on model feedback, repeatedly submitting queries with detectable malicious instructions during the attack search process. Although these approaches are effective, the attacks may be intercepted by content moderators during the search process. We propose an improved transfer attack method that guides malicious prompt construction by locally training a mirror model of the target black-box model through benign data distillation. This method offers enhanced stealth, as it does not involve submitting identifiable malicious instructions to the target model during the search phase. Our approach achieved a maximum attack success rate of 92%, or a balanced value of 80% with an average of 1.5 detectable jailbreak queries per sample against GPT-3.5 Turbo on a subset of AdvBench. These results underscore the need for more robust defense mechanisms.
comment: Accepted by NAACL 2025
♻ ☆ SRAG: Structured Retrieval-Augmented Generation for Multi-Entity Question Answering over Wikipedia Graph
Multi-entity question answering (MEQA) poses significant challenges for large language models (LLMs), which often struggle to consolidate scattered information across multiple documents. An example question might be "What is the distribution of IEEE Fellows among various fields of study?", which requires retrieving information from diverse sources e.g., Wikipedia pages. The effectiveness of current retrieval-augmented generation (RAG) methods is limited by the LLMs' capacity to aggregate insights from numerous pages. To address this gap, this paper introduces a structured RAG (SRAG) framework that systematically organizes extracted entities into relational tables (e.g., tabulating entities with schema columns like "name" and "field of study") and then apply table-based reasoning techniques. Our approach decouples retrieval and reasoning, enabling LLMs to focus on structured data analysis rather than raw text aggregation. Extensive experiments on Wikipedia-based multi-entity QA tasks demonstrate that SRAG significantly outperforms state-of-the-art long-context LLMs and RAG solutions, achieving a 29.6% improvement in accuracy. The results underscore the efficacy of structuring unstructured data to enhance LLMs' reasoning capabilities.
♻ ☆ HelpSteer2-Preference: Complementing Ratings with Preferences ICLR 2025
Reward models are critical for aligning models to follow instructions, and are typically trained following one of two popular paradigms: Bradley-Terry style or Regression style. However, there is a lack of evidence that either approach is better than the other, when adequately matched for data. This is primarily because these approaches require data collected in different (but incompatible) formats, meaning that adequately matched data is not available in existing public datasets. To tackle this problem, we release preference annotations (designed for Bradley-Terry training) to complement existing ratings (designed for Regression style training) in the HelpSteer2 dataset. To improve data interpretability, preference annotations are accompanied with human-written justifications. Using this data, we conduct the first head-to-head comparison of Bradley-Terry and Regression models when adequately matched for data. Based on insights derived from such a comparison, we propose a novel approach to combine Bradley-Terry and Regression reward modeling. A Llama-3.1-70B-Instruct model tuned with this approach scores 94.1 on RewardBench, emerging top of more than 140 reward models as of 1 Oct 2024. This reward model can then be used with REINFORCE algorithm (RLHF) to align an Instruct model to reach 85.0 on Arena Hard, which is No. 1 as of 1 Oct 2024. We open-source this dataset (CC-BY-4.0 license) at https://huggingface.co/datasets/nvidia/HelpSteer2#preferences-new -- 1-oct-2024 and openly release the trained Reward and Instruct models at https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Reward and https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct
comment: Accepted to ICLR 2025; 28 pages, 3 figures
♻ ☆ Women, Infamous, and Exotic Beings: What Honorific Usages in Wikipedia Reveal about the Socio-Cultural Norms
Honorifics serve as powerful linguistic markers that reflect social hierarchies and cultural values. This paper presents a large-scale, cross-linguistic exploration of usage of honorific pronouns in Bengali and Hindi Wikipedia articles, shedding light on how socio-cultural factors shape language. Using LLM (GPT-4o), we annotated 10, 000 articles of real and fictional beings in each language for several sociodemographic features such as gender, age, fame, and exoticness, and the use of honorifics. We find that across all feature combinations, use of honorifics is consistently more common in Bengali than Hindi. For both languages, the use non-honorific pronouns is more commonly observed for infamous, juvenile, and exotic beings. Notably, we observe a gender bias in use of honorifics in Hindi, with men being more commonly referred to with honorifics than women.
♻ ☆ Weak-to-Strong Preference Optimization: Stealing Reward from Weak Aligned Model ICLR 2025
Aligning language models (LMs) with human preferences has become a key area of research, enabling these models to meet diverse user needs better. Inspired by weak-to-strong generalization, where a strong LM fine-tuned on labels generated by a weaker model can consistently outperform its weak supervisor, we extend this idea to model alignment. In this work, we observe that the alignment behavior in weaker models can be effectively transferred to stronger models and even exhibit an amplification effect. Based on this insight, we propose a method called Weak-to-Strong Preference Optimization (WSPO), which achieves strong model alignment by learning the distribution differences before and after the alignment of the weak model. Experiments demonstrate that WSPO delivers outstanding performance, improving the win rate of Qwen2-7B-Instruct on Arena-Hard from 39.70 to 49.60, achieving a remarkable 47.04 length-controlled win rate on AlpacaEval 2, and scoring 7.33 on MT-bench. Our results suggest that using the weak model to elicit a strong model with a high alignment ability is feasible.
comment: ICLR 2025(Spotlight)
♻ ☆ Social Genome: Grounded Social Reasoning Abilities of Multimodal Models
Social reasoning abilities are crucial for AI systems to effectively interpret and respond to multimodal human communication and interaction within social contexts. We introduce Social Genome, the first benchmark for fine-grained, grounded social reasoning abilities of multimodal models. Social Genome contains 272 videos of interactions and 1,486 human-annotated reasoning traces related to inferences about these interactions. These traces contain 5,777 reasoning steps that reference evidence from visual cues, verbal cues, vocal cues, and external knowledge (contextual knowledge external to videos). Social Genome is also the first modeling challenge to study external knowledge in social reasoning. Social Genome computes metrics to holistically evaluate semantic and structural qualities of model-generated social reasoning traces. We demonstrate the utility of Social Genome through experiments with state-of-the-art models, identifying performance gaps and opportunities for future research to improve the grounded social reasoning abilities of multimodal models.
comment: Under Review, 22 pages
♻ ☆ When Claims Evolve: Evaluating and Enhancing the Robustness of Embedding Models Against Misinformation Edits
Online misinformation remains a critical challenge, and fact-checkers increasingly rely on embedding-based methods to retrieve relevant fact-checks. Yet, when debunked claims reappear in edited forms, the performance of these methods is unclear. In this work, we introduce a taxonomy of six common real-world misinformation edits and propose a perturbation framework that generates valid, natural claim variations. Our multi-stage retrieval evaluation reveals that standard embedding models struggle with user-introduced edits, while LLM-distilled embeddings offer improved robustness at a higher computational cost. Although a strong reranker helps mitigate some issues, it cannot fully compensate for first-stage retrieval gaps. Addressing these retrieval gaps, our train- and inference-time mitigation approaches enhance in-domain robustness by up to 17 percentage points and boost out-of-domain generalization by 10 percentage points over baseline models. Overall, our findings provide practical improvements to claim-matching systems, enabling more reliable fact-checking of evolving misinformation.
♻ ☆ Semi-Parametric Retrieval via Binary Bag-of-Tokens Index
Information retrieval has transitioned from standalone systems into essential components across broader applications, with indexing efficiency, cost-effectiveness, and freshness becoming increasingly critical yet often overlooked. In this paper, we introduce SemI-parametric Disentangled Retrieval (SiDR), a bi-encoder retrieval framework that decouples retrieval index from neural parameters to enable efficient, low-cost, and parameter-agnostic indexing for emerging use cases. Specifically, in addition to using embeddings as indexes like existing neural retrieval methods, SiDR supports a non-parametric tokenization index for search, achieving BM25-like indexing complexity with significantly better effectiveness. Our comprehensive evaluation across 16 retrieval benchmarks demonstrates that SiDR outperforms both neural and term-based retrieval baselines under the same indexing workload: (i) When using an embedding-based index, SiDR exceeds the performance of conventional neural retrievers while maintaining similar training complexity; (ii) When using a tokenization-based index, SiDR drastically reduces indexing cost and time, matching the complexity of traditional term-based retrieval, while consistently outperforming BM25 on all in-domain datasets; (iii) Additionally, we introduce a late parametric mechanism that matches BM25 index preparation time while outperforming other neural retrieval baselines in effectiveness.
♻ ☆ Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed Transformer
We propose Union-of-Experts (UoE), which decomposes transformer into an equitant group of experts, and then implement selective routing on input data and experts. Our approach advances MoE design with four key innovations: (1) We conducted equitant expert decomposition on both MLP blocks and attention blocks based on matrix partition in tensor parallelism. (2) We developed two routing paradigms: patch-wise data selection and expert selection, to apply routing across different levels. (3) We design the architecture of UoE model, including Selective Multi-Head Attention (SMHA) and Union-of-MLP-Experts (UoME). (4) We develop parallel implementation of UoE's routing and computation operation, and optimize efficiency based on the hardware processing analysis. The experiments demonstrate that the UoE model surpass Full Attention, state-of-art MoEs and efficient transformers (including the model architecture of recently proposed DeepSeek-V3) in several tasks across image and natural language domains. In language modeling tasks, we achieve an average reduction of 2.38 in perplexity compared to the best-performed MoE method with an average of 76% FLOPs. In Long Range Arena benchmark, we recorded an average score that is at least 0.68% higher than all comparison models including Full Attention, MoEs, and transformer variants, with only 50% FLOPs of the best MoE method. In image classification, our model yielded an average accuracy improvement of 1.75% than the best model while maintaining comparable FLOPs. The source codes are available at https://github.com/YujiaoYang-work/UoE.
comment: 17 pages
♻ ☆ Pap2Pat: Benchmarking Outline-Guided Long-Text Patent Generation with Patent-Paper Pairs
Dealing with long and highly complex technical text is a challenge for Large Language Models (LLMs), which still have to unfold their potential in supporting expensive and timeintensive processes like patent drafting. Within patents, the description constitutes more than 90% of the document on average. Yet, its automatic generation remains understudied. When drafting patent applications, patent attorneys typically receive invention reports (IRs), which are usually confidential, hindering research on LLM-supported patent drafting. Often, prepublication research papers serve as IRs. We leverage this duality to build PAP2PAT, an open and realistic benchmark for patent drafting consisting of 1.8k patent-paper pairs describing the same inventions. To address the complex longdocument patent generation task, we propose chunk-based outline-guided generation using the research paper as invention specification. Our extensive evaluation using PAP2PAT and a human case study show that LLMs can effectively leverage information from the paper, but still struggle to provide the necessary level of detail. Fine-tuning leads to more patent-style language, but also to more hallucination. We release our data and code https://github.com/boschresearch/Pap2Pat.
♻ ☆ Measuring Human and AI Values Based on Generative Psychometrics with Large Language Models AAAI 2025
Human values and their measurement are long-standing interdisciplinary inquiry. Recent advances in AI have sparked renewed interest in this area, with large language models (LLMs) emerging as both tools and subjects of value measurement. This work introduces Generative Psychometrics for Values (GPV), an LLM-based, data-driven value measurement paradigm, theoretically grounded in text-revealed selective perceptions. The core idea is to dynamically parse unstructured texts into perceptions akin to static stimuli in traditional psychometrics, measure the value orientations they reveal, and aggregate the results. Applying GPV to human-authored blogs, we demonstrate its stability, validity, and superiority over prior psychological tools. Then, extending GPV to LLM value measurement, we advance the current art with 1) a psychometric methodology that measures LLM values based on their scalable and free-form outputs, enabling context-specific measurement; 2) a comparative analysis of measurement paradigms, indicating response biases of prior methods; and 3) an attempt to bridge LLM values and their safety, revealing the predictive power of different value systems and the impacts of various values on LLM safety. Through interdisciplinary efforts, we aim to leverage AI for next-generation psychometrics and psychometrics for value-aligned AI.
comment: Accepted at AAAI 2025
♻ ☆ Autoformalizing Natural Language to First-Order Logic: A Case Study in Logical Fallacy Detection
Translating natural language into formal language such as First-Order Logic (FOL) is a foundational challenge in NLP with wide-ranging applications in automated reasoning, misinformation tracking, and knowledge validation. In this paper, we introduce Natural Language to First-Order Logic (NL2FOL), a framework to autoformalize natural language to FOL step by step using Large Language Models (LLMs). Our approach addresses key challenges in this translation process, including the integration of implicit background knowledge. By leveraging structured representations generated by NL2FOL, we use Satisfiability Modulo Theory (SMT) solvers to reason about the logical validity of natural language statements. We present logical fallacy detection as a case study to evaluate the efficacy of NL2FOL. Being neurosymbolic, our approach also provides interpretable insights into the reasoning process and demonstrates robustness without requiring model fine-tuning or labeled training data. Our framework achieves strong performance on multiple datasets. On the LOGIC dataset, NL2FOL achieves an F1-score of 78%, while generalizing effectively to the LOGICCLIMATE dataset with an F1-score of 80%.
♻ ☆ An LLM-based Agent for Reliable Docker Environment Configuration
Environment configuration is a critical yet time-consuming step in software development, especially when dealing with unfamiliar code repositories. While Large Language Models (LLMs) demonstrate the potential to accomplish software engineering tasks, existing methods for environment configuration often rely on manual efforts or fragile scripts, leading to inefficiencies and unreliable outcomes. We introduce Repo2Run, the first LLM-based agent designed to fully automate environment configuration and generate executable Dockerfiles for arbitrary Python repositories. We address two major challenges: (1) enabling the LLM agent to configure environments within isolated Docker containers, and (2) ensuring the successful configuration process is recorded and accurately transferred to a Dockerfile without error. To achieve this, we propose atomic configuration synthesis, featuring a dual-environment architecture (internal and external environment) with a rollback mechanism to prevent environment "pollution" from failed commands, guaranteeing atomic execution (execute fully or not at all) and a Dockerfile generator to transfer successful configuration steps into runnable Dockerfiles. We evaluate Repo2Run~on our proposed benchmark of 420 recent Python repositories with unit tests, where it achieves an 86.0% success rate, outperforming the best baseline by 63.9%. Repo2Run is available at https://github.com/bytedance/Repo2Run.
♻ ☆ Learning to Generate Structured Output with Schema Reinforcement Learning
This study investigates the structured generation capabilities of large language models (LLMs), focusing on producing valid JSON outputs against a given schema. Despite the widespread use of JSON in integrating language models with programs, there is a lack of comprehensive analysis and benchmarking of these capabilities. We explore various aspects of JSON generation, such as structure understanding, escaping, and natural language description, to determine how to assess and enable LLMs to generate valid responses. Building upon this, we propose SchemaBench features around 40K different JSON schemas to obtain and assess models' abilities in generating valid JSON. We find that the latest LLMs are still struggling to generate a valid JSON string. Moreover, we demonstrate that incorporating reinforcement learning with a Fine-grained Schema Validator can further enhance models' understanding of JSON schema, leading to improved performance. Our models demonstrate significant improvement in both generating JSON outputs and downstream tasks.
comment: 8 pages, 4 figures
♻ ☆ Dual Reasoning: A GNN-LLM Collaborative Framework for Knowledge Graph Question Answering
Large Language Models (LLMs) excel at intuitive, implicit reasoning. Guiding LLMs to construct thought chains can enhance their deliberate reasoning abilities, but also faces challenges such as hallucination. Knowledge Graphs (KGs) can provide explicit structured knowledge for LLMs to alleviate these issues. However, existing KG-enhanced methods often overlook explicit graph learning, making it challenging to efficiently provide precise reasoning chains for LLMs. Following dual-process theory, we propose Dual-Reasoning (DualR), a novel framework that integrates an external system based on Graph Neural Network (GNN) for explicit reasoning on KGs, complementing the implicit reasoning of LLMs through externalized reasoning chains. DualR designs an LLM-empowered GNN module for explicit learning on KGs, efficiently extracting high-quality reasoning chains. These reasoning chains are then refined to a knowledge-enhanced multiple-choice prompt, guiding a frozen LLM to reason thoughtfully for final answer determination. Extensive experiments on three benchmark KGQA datasets demonstrate that DualR achieves state-of-the-art performance while maintaining high efficiency and interpretability.
♻ ☆ R2-KG: General-Purpose Dual-Agent Framework for Reliable Reasoning on Knowledge Graphs
Recent studies have combined Large Language Models (LLMs) with Knowledge Graphs (KGs) to enhance reasoning, improving inference accuracy without additional training while mitigating hallucination. However, existing frameworks are often rigid, struggling to adapt to KG or task changes. They also rely heavily on powerful LLMs for reliable (i.e., trustworthy) reasoning. To address this, We introduce R2-KG, a plug-and-play, dual-agent framework that separates reasoning into two roles: an Operator (a low-capacity LLM) that gathers evidence and a Supervisor (a high-capacity LLM) that makes final judgments. This design is cost-efficient for LLM inference while still maintaining strong reasoning accuracy. Additionally, R2-KG employs an Abstention mechanism, generating answers only when sufficient evidence is collected from KG, which significantly enhances reliability. Experiments across multiple KG-based reasoning tasks show that R2-KG consistently outperforms baselines in both accuracy and reliability, regardless of the inherent capability of LLMs used as the Operator. Further experiments reveal that the single-agent version of R2-KG, equipped with a strict self-consistency strategy, achieves significantly higher-than-baseline reliability while reducing inference cost. However, it also leads to a higher abstention rate in complex KGs. Our findings establish R2-KG as a flexible and cost-effective solution for KG-based reasoning. It reduces reliance on high-capacity LLMs while ensuring trustworthy inference.
♻ ☆ Markov Chain of Thought for Efficient Mathematical Reasoning NAACL 2025
Chain of Thought (CoT) of multi-step benefits from the logical structure of the reasoning steps and task-specific actions, significantly enhancing the mathematical reasoning capabilities of large language models. As the prevalence of long CoT, the number of reasoning steps exceeds manageable token limits and leads to higher computational demands. Inspired by the fundamental logic of human cognition, "derive, then reduce", we conceptualize the standard multi-step CoT as a novel Markov Chain of Thought (MCoT). In this study, we consider the mathematical reasoning task, defining each reasoning step as text accompanied by a Python code snippet. To facilitate a longer reasoning path, self-correction is enabled through interactions with the code interpreter. Our MCoT aims to compress previous reasoning steps into a simplified question, enabling efficient next-step inference without relying on a lengthy KV cache. In our experiments, we curate the $\texttt{MCoTInstruct}$ dataset, and the empirical results indicate that MCoT not only significantly enhances efficiency but also maintains comparable accuracy. While much remains to be explored, this work paves the way for exploring the long CoT reasoning abilities of LLMs. The code is available at https://github.com/james-yw/Markov-Chain-of-Thought
comment: Camera ready version for NAACL 2025 Main
♻ ☆ Investigating Non-Transitivity in LLM-as-a-Judge
Automatic evaluation methods based on large language models (LLMs) are emerging as the standard tool for assessing the instruction-following abilities of LLM-based agents. The most common method in this paradigm, pairwise comparisons with a baseline model, critically depends on the assumption of transitive preferences. However, the validity of this assumption remains largely unexplored. In this study, we investigate the presence of non-transitivity within the AlpacaEval framework and analyze its effects on model rankings. We find that LLM judges exhibit non-transitive preferences, leading to rankings that are sensitive to the choice of the baseline model. To mitigate this issue, we show that round-robin tournaments combined with Bradley-Terry models of preference can produce more reliable rankings. Notably, our method increases both the Spearman correlation and the Kendall correlation with Chatbot Arena (95.0% -> 96.4% and 82.1% -> 86.3% respectively). To address the computational cost of round-robin tournaments, we propose Swiss-Wise Iterative Matchmaking (Swim) tournaments, using a dynamic matching strategy to capture the benefits of round-robin tournaments while maintaining computational efficiency.
comment: 8 pages, 6 figures, 2 tables (30 pages, 11 figures, 8 tables including references and appendices)
♻ ☆ Training and Evaluating Language Models with Template-based Data Generation
The rapid advancement of large language models (LLMs) such as GPT-3, PaLM, and Llama has significantly transformed natural language processing, showcasing remarkable capabilities in understanding and generating language. However, these models often struggle with tasks requiring complex reasoning, particularly in mathematical problem-solving, due in part to the scarcity of large-scale, high-quality, domain-specific datasets necessary for training sophisticated reasoning abilities. To address this limitation, we introduce Template-based Data Generation (TDG), a novel approach that leverages LLMs (GPT-4) to automatically generate parameterized meta-templates, which are then used to synthesize a vast array of high-quality problems and solutions. Leveraging TDG, we create TemplateMath Part I: TemplateGSM, a dataset comprising over 7 million synthetically generated grade school math problems--each accompanied by code-based and natural language solutions--with the potential to generate an effectively unlimited number more. This dataset alleviates the scarcity of large-scale mathematical datasets and serves as a valuable resource for pre-training, fine-tuning, and evaluating LLMs in mathematical reasoning. Our method not only enables the generation of virtually infinite data but also elevates data augmentation to a new level by using GPT-4 for meta-template generation, ensuring diverse and high-quality problem structures. The TemplateMath Part I: TemplateGSM dataset is publicly available at https://huggingface.co/datasets/math-ai/TemplateGSM. The code is available at https://github.com/iiis-ai/TemplateMath.
comment: 9 pages, 2 figures
♻ ☆ Legal Fact Prediction: The Missing Piece in Legal Judgment Prediction
Legal judgment prediction (LJP), which enables litigants and their lawyers to forecast judgment outcomes and refine litigation strategies, has emerged as a crucial legal NLP task. Existing studies typically utilize legal facts, i.e., facts that have been established by evidence and determined by the judge, to predict the judgment. However, legal facts are often difficult to obtain in the early stages of litigation, significantly limiting the practical applicability of fact-based LJP. To address this limitation, we propose a novel legal NLP task: \textit{legal fact prediction} (LFP), which takes the evidence submitted by litigants for trial as input to predict legal facts, thereby empowering fact-based LJP technologies to perform prediction in the absence of ground-truth legal facts. We also propose the first benchmark dataset, LFPBench, for evaluating the LFP task. Our extensive experiments on LFPBench demonstrate the effectiveness of LFP-empowered LJP and highlight promising research directions for LFP. Our code and data are available at https://github.com/HPRCEST/LFPBench.
♻ ☆ Prompting with Phonemes: Enhancing LLMs' Multilinguality for Non-Latin Script Languages NAACL 2025
Although multilingual LLMs have achieved remarkable performance across benchmarks, we find they continue to underperform on non-Latin script languages across contemporary LLM families. This discrepancy arises from the fact that LLMs are pretrained with orthographic scripts, which are dominated by Latin characters that obscure their shared phonology with non-Latin scripts. We propose leveraging phonemic transcriptions as complementary signals to induce script-invariant representations. Our study demonstrates that integrating phonemic signals improves performance across both non-Latin and Latin script languages, with a particularly significant impact on closing the performance gap between the two. Through detailed experiments, we show that phonemic and orthographic scripts retrieve distinct examples for in-context learning (ICL). This motivates our proposed Mixed-ICL retrieval strategy, where further aggregation from both leads to our significant performance improvements for both Latin script languages (up to 12.6%) and non-Latin script languages (up to 15.1%) compared to randomized ICL retrieval.
comment: Accepted for NAACL 2025 (Main Conference)
♻ ☆ GENERator: A Long-Context Generative Genomic Foundation Model
Advancements in DNA sequencing technologies have significantly improved our ability to decode genomic sequences. However, the prediction and interpretation of these sequences remain challenging due to the intricate nature of genetic material. Large language models (LLMs) have introduced new opportunities for biological sequence analysis. Recent developments in genomic language models have underscored the potential of LLMs in deciphering DNA sequences. Nonetheless, existing models often face limitations in robustness and application scope, primarily due to constraints in model structure and training data scale. To address these limitations, we present GENERator, a generative genomic foundation model featuring a context length of 98k base pairs (bp) and 1.2B parameters. Trained on an expansive dataset comprising 386B bp of eukaryotic DNA, the GENERator demonstrates state-of-the-art performance across both established and newly proposed benchmarks. The model adheres to the central dogma of molecular biology, accurately generating protein-coding sequences that translate into proteins structurally analogous to known families. It also shows significant promise in sequence optimization, particularly through the prompt-responsive generation of enhancer sequences with specific activity profiles. These capabilities position the GENERator as a pivotal tool for genomic research and biotechnological advancement, enhancing our ability to interpret and predict complex biological systems and enabling precise genomic interventions. Implementation details and supplementary resources are available at https://github.com/GenerTeam/GENERator.
♻ ☆ Unraveling and Mitigating Retriever Inconsistencies in Retrieval-Augmented Large Language Models ACL 2024
Although Retrieval-Augmented Large Language Models (RALMs) demonstrate their superiority in terms of factuality, they do not consistently outperform the original retrieval-free Language Models (LMs). Our experiments reveal that this example-level performance inconsistency exists not only between retrieval-augmented and retrieval-free LM but also among different retrievers. To understand this phenomenon, we investigate the degeneration behavior of RALMs and theoretically decompose it into four categories. Further analysis based on our decomposition reveals that the innate difference in knowledge sources and the unpredictable degeneration of the reader model contribute most to the inconsistency. Drawing from our analysis, we introduce Ensemble of Retrievers (EoR), a trainable framework that can adaptively retrieve from different knowledge sources and effectively decrease unpredictable reader errors. Our experiments on Open Domain Question Answering show that EoR substantially improves performance over the RALM with a single retriever by considerably reducing inconsistent behaviors.
comment: ACL 2024 (findings)
♻ ☆ MultiChallenge: A Realistic Multi-Turn Conversation Evaluation Benchmark Challenging to Frontier LLMs
We present MultiChallenge, a pioneering benchmark evaluating large language models (LLMs) on conducting multi-turn conversations with human users, a crucial yet underexamined capability for their applications. MultiChallenge identifies four categories of challenges in multi-turn conversations that are not only common and realistic among current human-LLM interactions, but are also challenging to all current frontier LLMs. All 4 challenges require accurate instruction-following, context allocation, and in-context reasoning at the same time. We also develop LLM as judge with instance-level rubrics to facilitate an automatic evaluation method with fair agreement with experienced human raters. Despite achieving near-perfect scores on existing multi-turn evaluation benchmarks, all frontier models have less than 50% accuracy on MultiChallenge, with the top-performing Claude 3.5 Sonnet (June 2024) achieving just a 41.4% average accuracy.
♻ ☆ Visual Description Grounding Reduces Hallucinations and Boosts Reasoning in LVLMs ICLR 2025
Large Vision-Language Models (LVLMs) often produce responses that misalign with factual information, a phenomenon known as hallucinations. While hallucinations are well-studied, the exact causes behind them remain underexplored. In this paper, we first investigate the root causes of hallucinations in LVLMs. Our findings reveal that existing mitigation techniques primarily reduce hallucinations for visual recognition prompts-those that require simple descriptions of visual elements-but fail for cognitive prompts that demand deliberate reasoning. We identify the core issue as a lack of true visual perception in LVLMs: although they can accurately recognize visual elements, they struggle to fully interpret these elements in the context of the input prompt and effectively link this recognition to their internal knowledge, which is critical for reasoning. To address this gap, we introduce Visual Description Grounded Decoding (VDGD), a simple, robust, and training-free method designed to enhance visual perception and improve reasoning capabilities in LVLMs. VDGD works by first generating a detailed description of the image and appending it as a prefix to the instruction. During response generation, tokens are sampled based on their KL divergence to the description, favoring candidates with lower divergence. Experimental results on multiple visual reasoning benchmarks and LVLMs demonstrate that VDGD consistently outperforms existing baselines 2% - 33%. Finally, we introduce VaLLu, a benchmark designed for comprehensive evaluation of the cognitive capabilities of LVLMs.
comment: Accepted to ICLR 2025. Project: https://sreyan88.github.io/VDGD/
♻ ☆ EchoQA: A Large Collection of Instruction Tuning Data for Echocardiogram Reports NeurIPS
We introduce a novel question-answering (QA) dataset using echocardiogram reports sourced from the Medical Information Mart for Intensive Care database. This dataset is specifically designed to enhance QA systems in cardiology, consisting of 771,244 QA pairs addressing a wide array of cardiac abnormalities and their severity. We compare large language models (LLMs), including open-source and biomedical-specific models for zero-shot evaluation, and closed-source models for zero-shot and three-shot evaluation. Our results show that fine-tuning LLMs improves performance across various QA metrics, validating the value of our dataset. Clinicians also qualitatively evaluate the best-performing model to assess the LLM responses for correctness. Further, we conduct fine-grained fairness audits to assess the bias-performance trade-off of LLMs across various social determinants of health. Our objective is to propel the field forward by establishing a benchmark for LLM AI agents aimed at supporting clinicians with cardiac differential diagnoses, thereby reducing the documentation burden that contributes to clinician burnout and enabling healthcare professionals to focus more on patient care.
comment: NeurIPS SafeGenAI 2024
♻ ☆ DoraemonGPT: Toward Understanding Dynamic Scenes with Large Language Models (Exemplified as A Video Agent)
Recent LLM-driven visual agents mainly focus on solving image-based tasks, which limits their ability to understand dynamic scenes, making it far from real-life applications like guiding students in laboratory experiments and identifying their mistakes. Hence, this paper explores DoraemonGPT, a comprehensive and conceptually elegant system driven by LLMs to understand dynamic scenes. Considering the video modality better reflects the ever-changing nature of real-world scenarios, we exemplify DoraemonGPT as a video agent. Given a video with a question/task, DoraemonGPT begins by converting the input video into a symbolic memory that stores task-related attributes. This structured representation allows for spatial-temporal querying and reasoning by well-designed sub-task tools, resulting in concise intermediate results. Recognizing that LLMs have limited internal knowledge when it comes to specialized domains (e.g., analyzing the scientific principles underlying experiments), we incorporate plug-and-play tools to assess external knowledge and address tasks across different domains. Moreover, a novel LLM-driven planner based on Monte Carlo Tree Search is introduced to explore the large planning space for scheduling various tools. The planner iteratively finds feasible solutions by backpropagating the result's reward, and multiple solutions can be summarized into an improved final answer. We extensively evaluate DoraemonGPT's effectiveness on three benchmarks and several in-the-wild scenarios. The code will be released at https://github.com/z-x-yang/DoraemonGPT.
♻ ☆ SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration ICLR 2025
Speculative decoding (SD) has emerged as a widely used paradigm to accelerate LLM inference without compromising quality. It works by first employing a compact model to draft multiple tokens efficiently and then using the target LLM to verify them in parallel. While this technique has achieved notable speedups, most existing approaches necessitate either additional parameters or extensive training to construct effective draft models, thereby restricting their applicability across different LLMs and tasks. To address this limitation, we explore a novel plug-and-play SD solution with layer-skipping, which skips intermediate layers of the target LLM as the compact draft model. Our analysis reveals that LLMs exhibit great potential for self-acceleration through layer sparsity and the task-specific nature of this sparsity. Building on these insights, we introduce SWIFT, an on-the-fly self-speculative decoding algorithm that adaptively selects intermediate layers of LLMs to skip during inference. SWIFT does not require auxiliary models or additional training, making it a plug-and-play solution for accelerating LLM inference across diverse input data streams. Our extensive experiments across a wide range of models and downstream tasks demonstrate that SWIFT can achieve over a 1.3x-1.6x speedup while preserving the original distribution of the generated text. We release our code in https://github.com/hemingkx/SWIFT.
comment: ICLR 2025, camera-ready version
♻ ☆ Unifying Multitrack Music Arrangement via Reconstruction Fine-Tuning and Efficient Tokenization IJCAI 2025
Automatic music arrangement streamlines the creation of musical variants for composers and arrangers, reducing reliance on extensive music expertise. However, existing methods suffer from inefficient tokenization, underutilization of pre-trained music language models (LMs), and suboptimal fidelity and coherence in generated arrangements. This paper introduces an efficient multitrack music tokenizer for unconditional and conditional symbolic music generation, along with a unified sequence-to-sequence reconstruction fine-tuning objective for pre-trained music LMs that balances task-specific needs with coherence constraints. Our approach achieves state-of-the-art results on band arrangement, piano reduction, and drum arrangement, surpassing task-specific models in both objective metrics and perceptual quality. Additionally, we demonstrate that generative pretraining significantly contributes to the performance across these arrangement tasks, especially when handling long segments with complex alignment.
comment: Submitted to IJCAI 2025
♻ ☆ Automatically Labeling Clinical Trial Outcomes: A Large-Scale Benchmark for Drug Development
Background The cost of drug discovery and development is substantial, with clinical trial outcomes playing a critical role in regulatory approval and patient care. However, access to large-scale, high-quality clinical trial outcome data remains limited, hindering advancements in predictive modeling and evidence-based decision-making. Methods We present the Clinical Trial Outcome (CTO) benchmark, a fully reproducible, large-scale repository encompassing approximately 125,000 drug and biologics trials. CTO integrates large language model (LLM) interpretations of publications, trial phase progression tracking, sentiment analysis from news sources, stock price movements of trial sponsors, and additional trial-related metrics. Furthermore, we manually annotated a dataset of clinical trials conducted between 2020 and 2024 to enhance the quality and reliability of outcome labels. Results The trial outcome labels in the CTO benchmark agree strongly with expert annotations, achieving an F1 score of 94 for Phase 3 trials and 91 across all phases. Additionally, benchmarking standard machine learning models on our manually annotated dataset revealed distribution shifts in recent trials, underscoring the necessity of continuously updated labeling approaches. Conclusions By analyzing CTO's performance on recent clinical trials, we demonstrate the ongoing need for high-quality, up-to-date trial outcome labels. We publicly release the CTO knowledge base and annotated labels at https://chufangao.github.io/CTOD, with regular updates to support research on clinical trial outcomes and inform data-driven improvements in drug development.
♻ ☆ EgoNormia: Benchmarking Physical Social Norm Understanding
Human activity is moderated by norms. However, machines are often trained without explicit supervision on norm understanding and reasoning, especially when the norms are grounded in a physical and social context. To improve and evaluate the normative reasoning capability of vision-language models (VLMs), we present EgoNormia $\|\epsilon\|$, consisting of 1,853 ego-centric videos of human interactions, each of which has two related questions evaluating both the prediction and justification of normative actions. The normative actions encompass seven categories: safety, privacy, proxemics, politeness, cooperation, coordination/proactivity, and communication/legibility. To compile this dataset at scale, we propose a novel pipeline leveraging video sampling, automatic answer generation, filtering, and human validation. Our work demonstrates that current state-of-the-art vision-language models lack robust norm understanding, scoring a maximum of 45% on EgoNormia (versus a human bench of 92%). Our analysis of performance in each dimension highlights the significant risks of safety, privacy, and the lack of collaboration and communication capability when applied to real-world agents. We additionally show that through a retrieval-based generation method, it is possible to use EgoNormia to enhance normative reasoning in VLMs.
♻ ☆ Adversarial Decoding: Generating Readable Documents for Adversarial Objectives
We design, implement, and evaluate adversarial decoding, a new, generic text generation technique that produces readable documents for different adversarial objectives. Prior methods either produce easily detectable gibberish, or cannot handle objectives that include embedding similarity. In particular, they only work for direct attacks (such as jailbreaking) and cannot produce adversarial text for realistic indirect injection, e.g., documents that (1) are retrieved in RAG systems in response to broad classes of queries, and also (2) adversarially influence subsequent generation. We also show that fluency (low perplexity) is not sufficient to evade filtering. We measure the effectiveness of adversarial decoding for different objectives, including RAG poisoning, jailbreaking, and evasion of defensive filters, and demonstrate that it outperforms existing methods while producing readable adversarial documents.
♻ ☆ METAL: A Multi-Agent Framework for Chart Generation with Test-Time Scaling
Chart generation aims to generate code to produce charts satisfying the desired visual properties, e.g., texts, layout, color, and type. It has great potential to empower the automatic professional report generation in financial analysis, research presentation, education, and healthcare. In this work, we build a vision-language model (VLM) based multi-agent framework for effective automatic chart generation. Generating high-quality charts requires both strong visual design skills and precise coding capabilities that embed the desired visual properties into code. Such a complex multi-modal reasoning process is difficult for direct prompting of VLMs. To resolve these challenges, we propose METAL, a multi-agent framework that decomposes the task of chart generation into the iterative collaboration among specialized agents. METAL achieves 5.2% improvement over the current best result in the chart generation task. The METAL framework exhibits the phenomenon of test-time scaling: its performance increases monotonically as the logarithmic computational budget grows from 512 to 8192 tokens. In addition, we find that separating different modalities during the critique process of METAL boosts the self-correction capability of VLMs in the multimodal context.
♻ ☆ Towards Fully-Automated Materials Discovery via Large-Scale Synthesis Dataset and Expert-Level LLM-as-a-Judge
Materials synthesis is vital for innovations such as energy storage, catalysis, electronics, and biomedical devices. Yet, the process relies heavily on empirical, trial-and-error methods guided by expert intuition. Our work aims to support the materials science community by providing a practical, data-driven resource. We have curated a comprehensive dataset of 17K expert-verified synthesis recipes from open-access literature, which forms the basis of our newly developed benchmark, AlchemyBench. AlchemyBench offers an end-to-end framework that supports research in large language models applied to synthesis prediction. It encompasses key tasks, including raw materials and equipment prediction, synthesis procedure generation, and characterization outcome forecasting. We propose an LLM-as-a-Judge framework that leverages large language models for automated evaluation, demonstrating strong statistical agreement with expert assessments. Overall, our contributions offer a supportive foundation for exploring the capabilities of LLMs in predicting and guiding materials synthesis, ultimately paving the way for more efficient experimental design and accelerated innovation in materials science.
comment: under review
♻ ☆ Distributional Associations vs In-Context Reasoning: A Study of Feed-forward and Attention Layers ICLR 2025
Large language models have been successful at tasks involving basic forms of in-context reasoning, such as generating coherent language, as well as storing vast amounts of knowledge. At the core of the Transformer architecture behind such models are feed-forward and attention layers, which are often associated to knowledge and reasoning, respectively. In this paper, we study this distinction empirically and theoretically in a controlled synthetic setting where certain next-token predictions involve both distributional and in-context information. We find that feed-forward layers tend to learn simple distributional associations such as bigrams, while attention layers focus on in-context reasoning. Our theoretical analysis identifies the noise in the gradients as a key factor behind this discrepancy. Finally, we illustrate how similar disparities emerge in pre-trained models through ablations on the Pythia model family on simple reasoning tasks.
comment: ICLR 2025
♻ ☆ NaijaNLP: A Survey of Nigerian Low-Resource Languages
With over 500 languages in Nigeria, three languages -- Hausa, Yor\`ub\'a and Igbo -- spoken by over 175 million people, account for about 60% of the spoken languages. However, these languages are categorised as low-resource due to insufficient resources to support tasks in computational linguistics. Several research efforts and initiatives have been presented, however, a coherent understanding of the state of Natural Language Processing (NLP) - from grammatical formalisation to linguistic resources that support complex tasks such as language understanding and generation is lacking. This study presents the first comprehensive review of advancements in low-resource NLP (LR-NLP) research across the three major Nigerian languages (NaijaNLP). We quantitatively assess the available linguistic resources and identify key challenges. Although a growing body of literature addresses various NLP downstream tasks in Hausa, Igbo, and Yor\`ub\'a, only about 25.1% of the reviewed studies contribute new linguistic resources. This finding highlights a persistent reliance on repurposing existing data rather than generating novel, high-quality resources. Additionally, language-specific challenges, such as the accurate representation of diacritics, remain under-explored. To advance NaijaNLP and LR-NLP more broadly, we emphasise the need for intensified efforts in resource enrichment, comprehensive annotation, and the development of open collaborative initiatives.
comment: 35 pages, 2 figures, 4 tables
♻ ☆ Persuade Me if You Can: A Framework for Evaluating Persuasion Effectiveness and Susceptibility Among Large Language Models
Large Language Models (LLMs) demonstrate persuasive capabilities that rival human-level persuasion. While these capabilities can be used for social good, they also present risks of potential misuse. Moreover, LLMs' susceptibility to persuasion raises concerns about alignment with ethical principles. To study these dynamics, we introduce Persuade Me If You Can (PMIYC), an automated framework for evaluating persuasion through multi-agent interactions. Here, Persuader agents engage in multi-turn conversations with the Persuadee agents, allowing us to measure LLMs' persuasive effectiveness and their susceptibility to persuasion. We conduct comprehensive evaluations across diverse LLMs, ensuring each model is assessed against others in both subjective and misinformation contexts. We validate the efficacy of our framework through human evaluations and show alignment with prior work. PMIYC offers a scalable alternative to human annotation for studying persuasion in LLMs. Through PMIYC, we find that Llama-3.3-70B and GPT-4o exhibit similar persuasive effectiveness, outperforming Claude 3 Haiku by 30%. However, GPT-4o demonstrates over 50% greater resistance to persuasion for misinformation compared to Llama-3.3-70B. These findings provide empirical insights into the persuasive dynamics of LLMs and contribute to the development of safer AI systems.
♻ ☆ Self-correction is Not An Innate Capability in Large Language Models: A Case Study of Moral Self-correction
Though there has been intensive attention to the self-correction capability of Large Language Models (LLMs), conclusions regarding its effectiveness remain varied. In this paper, we investigate a fundamental question: is moral self-correction an innate capability in LLMs? To explore this, we conduct (1) a mechanistic analysis of how key components of self-correction, such as Chain-of-Thought (CoT) reasoning and external feedback, interact to enable moral self-correction; and (2) a behavioral analysis of LLMs' ability to distinguish between desired and undesired outputs, introducing a self-distinguish framework. Our mechanistic analysis reveals that LLMs struggle to effectively leverage helpful feedback, and conflicts can arise between feedback and CoT reasoning. These limitations suggest that LLMs fail to identify useful contextual information, instead prioritizing their own internal knowledge. Additionally, our behavioral analysis indicates that LLMs struggle to differentiate among their own outputs. Based on these empirical findings across two analytical dimensions, mechanism and behavior, we argue that moral self-correction is not an innate capability of LLMs.
♻ ☆ Shh, don't say that! Domain Certification in LLMs ICLR
Large language models (LLMs) are often deployed to perform constrained tasks, with narrow domains. For example, customer support bots can be built on top of LLMs, relying on their broad language understanding and capabilities to enhance performance. However, these LLMs are adversarially susceptible, potentially generating outputs outside the intended domain. To formalize, assess, and mitigate this risk, we introduce domain certification; a guarantee that accurately characterizes the out-of-domain behavior of language models. We then propose a simple yet effective approach, which we call VALID that provides adversarial bounds as a certificate. Finally, we evaluate our method across a diverse set of datasets, demonstrating that it yields meaningful certificates, which bound the probability of out-of-domain samples tightly with minimum penalty to refusal behavior.
comment: 10 pages, includes appendix Published in International Conference on Learning Representations (ICLR) 2025
♻ ☆ Large Language Model for Qualitative Research -- A Systematic Mapping Study ICSE
The exponential growth of text-based data in domains such as healthcare, education, and social sciences has outpaced the capacity of traditional qualitative analysis methods, which are time-intensive and prone to subjectivity. Large Language Models (LLMs), powered by advanced generative AI, have emerged as transformative tools capable of automating and enhancing qualitative analysis. This study systematically maps the literature on the use of LLMs for qualitative research, exploring their application contexts, configurations, methodologies, and evaluation metrics. Findings reveal that LLMs are utilized across diverse fields, demonstrating the potential to automate processes traditionally requiring extensive human input. However, challenges such as reliance on prompt engineering, occasional inaccuracies, and contextual limitations remain significant barriers. This research highlights opportunities for integrating LLMs with human expertise, improving model robustness, and refining evaluation methodologies. By synthesizing trends and identifying research gaps, this study aims to guide future innovations in the application of LLMs for qualitative analysis.
comment: 8 pages, includes 1 figures and 3 tables. Submitted and Accepted to the WSESE 2025 ICSE Workshop
Computer Vision and Pattern Recognition 164
☆ FluidNexus: 3D Fluid Reconstruction and Prediction from a Single Video CVPR 2025
We study reconstructing and predicting 3D fluid appearance and velocity from a single video. Current methods require multi-view videos for fluid reconstruction. We present FluidNexus, a novel framework that bridges video generation and physics simulation to tackle this task. Our key insight is to synthesize multiple novel-view videos as references for reconstruction. FluidNexus consists of two key components: (1) a novel-view video synthesizer that combines frame-wise view synthesis with video diffusion refinement for generating realistic videos, and (2) a physics-integrated particle representation coupling differentiable simulation and rendering to simultaneously facilitate 3D fluid reconstruction and prediction. To evaluate our approach, we collect two new real-world fluid datasets featuring textured backgrounds and object interactions. Our method enables dynamic novel view synthesis, future prediction, and interaction simulation from a single fluid video. Project website: https://yuegao.me/FluidNexus.
comment: CVPR 2025. Project website: https://yuegao.me/FluidNexus
☆ Floxels: Fast Unsupervised Voxel Based Scene Flow Estimation CVPR 2025
Scene flow estimation is a foundational task for many robotic applications, including robust dynamic object detection, automatic labeling, and sensor synchronization. Two types of approaches to the problem have evolved: 1) Supervised and 2) optimization-based methods. Supervised methods are fast during inference and achieve high-quality results, however, they are limited by the need for large amounts of labeled training data and are susceptible to domain gaps. In contrast, unsupervised test-time optimization methods do not face the problem of domain gaps but usually suffer from substantial runtime, exhibit artifacts, or fail to converge to the right solution. In this work, we mitigate several limitations of existing optimization-based methods. To this end, we 1) introduce a simple voxel grid-based model that improves over the standard MLP-based formulation in multiple dimensions and 2) introduce a new multiframe loss formulation. 3) We combine both contributions in our new method, termed Floxels. On the Argoverse 2 benchmark, Floxels is surpassed only by EulerFlow among unsupervised methods while achieving comparable performance at a fraction of the computational cost. Floxels achieves a massive speedup of more than ~60 - 140x over EulerFlow, reducing the runtime from a day to 10 minutes per sequence. Over the faster but low-quality baseline, NSFP, Floxels achieves a speedup of ~14x.
comment: Accepted at CVPR 2025
☆ Iris Style Transfer: Enhancing Iris Recognition with Style Features and Privacy Preservation through Neural Style Transfer
Iris texture is widely regarded as a gold standard biometric modality for authentication and identification. The demand for robust iris recognition methods, coupled with growing security and privacy concerns regarding iris attacks, has escalated recently. Inspired by neural style transfer, an advanced technique that leverages neural networks to separate content and style features, we hypothesize that iris texture's style features provide a reliable foundation for recognition and are more resilient to variations like rotation and perspective shifts than traditional approaches. Our experimental results support this hypothesis, showing a significantly higher classification accuracy compared to conventional features. Further, we propose using neural style transfer to mask identifiable iris style features, ensuring the protection of sensitive biometric information while maintaining the utility of eye images for tasks like eye segmentation and gaze estimation. This work opens new avenues for iris-oriented, secure, and privacy-aware biometric systems.
comment: 14 pages main paper, 4 pages appendix
☆ DEAL-YOLO: Drone-based Efficient Animal Localization using YOLO ICLR 2025
Although advances in deep learning and aerial surveillance technology are improving wildlife conservation efforts, complex and erratic environmental conditions still pose a problem, requiring innovative solutions for cost-effective small animal detection. This work introduces DEAL-YOLO, a novel approach that improves small object detection in Unmanned Aerial Vehicle (UAV) images by using multi-objective loss functions like Wise IoU (WIoU) and Normalized Wasserstein Distance (NWD), which prioritize pixels near the centre of the bounding box, ensuring smoother localization and reducing abrupt deviations. Additionally, the model is optimized through efficient feature extraction with Linear Deformable (LD) convolutions, enhancing accuracy while maintaining computational efficiency. The Scaled Sequence Feature Fusion (SSFF) module enhances object detection by effectively capturing inter-scale relationships, improving feature representation, and boosting metrics through optimized multiscale fusion. Comparison with baseline models reveals high efficacy with up to 69.5\% fewer parameters compared to vanilla Yolov8-N, highlighting the robustness of the proposed modifications. Through this approach, our paper aims to facilitate the detection of endangered species, animal population analysis, habitat monitoring, biodiversity research, and various other applications that enrich wildlife conservation efforts. DEAL-YOLO employs a two-stage inference paradigm for object detection, refining selected regions to improve localization and confidence. This approach enhances performance, especially for small instances with low objectness scores.
comment: Accepted as a Poster at the ML4RS Workshop at ICLR 2025
☆ Teach YOLO to Remember: A Self-Distillation Approach for Continual Object Detection
Real-time object detectors like YOLO achieve exceptional performance when trained on large datasets for multiple epochs. However, in real-world scenarios where data arrives incrementally, neural networks suffer from catastrophic forgetting, leading to a loss of previously learned knowledge. To address this, prior research has explored strategies for Class Incremental Learning (CIL) in Continual Learning for Object Detection (CLOD), with most approaches focusing on two-stage object detectors. However, existing work suggests that Learning without Forgetting (LwF) may be ineffective for one-stage anchor-free detectors like YOLO due to noisy regression outputs, which risk transferring corrupted knowledge. In this work, we introduce YOLO LwF, a self-distillation approach tailored for YOLO-based continual object detection. We demonstrate that when coupled with a replay memory, YOLO LwF significantly mitigates forgetting. Compared to previous approaches, it achieves state-of-the-art performance, improving mAP by +2.1% and +2.9% on the VOC and COCO benchmarks, respectively.
☆ What Are You Doing? A Closer Look at Controllable Human Video Generation
High-quality benchmarks are crucial for driving progress in machine learning research. However, despite the growing interest in video generation, there is no comprehensive dataset to evaluate human generation. Humans can perform a wide variety of actions and interactions, but existing datasets, like TikTok and TED-Talks, lack the diversity and complexity to fully capture the capabilities of video generation models. We close this gap by introducing `What Are You Doing?' (WYD): a new benchmark for fine-grained evaluation of controllable image-to-video generation of humans. WYD consists of 1{,}544 captioned videos that have been meticulously collected and annotated with 56 fine-grained categories. These allow us to systematically measure performance across 9 aspects of human generation, including actions, interactions and motion. We also propose and validate automatic metrics that leverage our annotations and better capture human evaluations. Equipped with our dataset and metrics, we perform in-depth analyses of seven state-of-the-art models in controllable image-to-video generation, showing how WYD provides novel insights about the capabilities of these models. We release our data and code to drive forward progress in human video generation modeling at https://github.com/google-deepmind/wyd-benchmark.
☆ Implicit Neural Representation for Video and Image Super-Resolution
We present a novel approach for super-resolution that utilizes implicit neural representation (INR) to effectively reconstruct and enhance low-resolution videos and images. By leveraging the capacity of neural networks to implicitly encode spatial and temporal features, our method facilitates high-resolution reconstruction using only low-resolution inputs and a 3D high-resolution grid. This results in an efficient solution for both image and video super-resolution. Our proposed method, SR-INR, maintains consistent details across frames and images, achieving impressive temporal stability without relying on the computationally intensive optical flow or motion estimation typically used in other video super-resolution techniques. The simplicity of our approach contrasts with the complexity of many existing methods, making it both effective and efficient. Experimental evaluations show that SR-INR delivers results on par with or superior to state-of-the-art super-resolution methods, while maintaining a more straightforward structure and reduced computational demands. These findings highlight the potential of implicit neural representations as a powerful tool for reconstructing high-quality, temporally consistent video and image signals from low-resolution data.
☆ RadIR: A Scalable Framework for Multi-Grained Medical Image Retrieval via Radiology Report Mining
Developing advanced medical imaging retrieval systems is challenging due to the varying definitions of `similar images' across different medical contexts. This challenge is compounded by the lack of large-scale, high-quality medical imaging retrieval datasets and benchmarks. In this paper, we propose a novel methodology that leverages dense radiology reports to define image-wise similarity ordering at multiple granularities in a scalable and fully automatic manner. Using this approach, we construct two comprehensive medical imaging retrieval datasets: MIMIC-IR for Chest X-rays and CTRATE-IR for CT scans, providing detailed image-image ranking annotations conditioned on diverse anatomical structures. Furthermore, we develop two retrieval systems, RadIR-CXR and model-ChestCT, which demonstrate superior performance in traditional image-image and image-report retrieval tasks. These systems also enable flexible, effective image retrieval conditioned on specific anatomical structures described in text, achieving state-of-the-art results on 77 out of 78 metrics.
☆ Transferable Foundation Models for Geometric Tasks on Point Cloud Representations: Geometric Neural Operators
We introduce methods for obtaining pretrained Geometric Neural Operators (GNPs) that can serve as basal foundation models for use in obtaining geometric features. These can be used within data processing pipelines for machine learning tasks and numerical methods. We show how our GNPs can be trained to learn robust latent representations for the differential geometry of point-clouds to provide estimates of metric, curvature, and other shape-related features. We demonstrate how our pre-trained GNPs can be used (i) to estimate the geometric properties of surfaces of arbitrary shape and topologies with robustness in the presence of noise, (ii) to approximate solutions of geometric partial differential equations (PDEs) on manifolds, and (iii) to solve equations for shape deformations such as curvature driven flows. We also release a package of the codes and weights for using our pre-trained GNPs for processing point cloud representations. This allows for incorporating our pre-trained GNPs as components for reuse within existing and new data processing pipelines. The GNPs also can be used as part of numerical solvers involving geometry or as part of methods for performing inference and other geometric tasks.
☆ Adaptive Prototype Learning for Multimodal Cancer Survival Analysis
Leveraging multimodal data, particularly the integration of whole-slide histology images (WSIs) and transcriptomic profiles, holds great promise for improving cancer survival prediction. However, excessive redundancy in multimodal data can degrade model performance. In this paper, we propose Adaptive Prototype Learning (APL), a novel and effective approach for multimodal cancer survival analysis. APL adaptively learns representative prototypes in a data-driven manner, reducing redundancy while preserving critical information. Our method employs two sets of learnable query vectors that serve as a bridge between high-dimensional representations and survival prediction, capturing task-relevant features. Additionally, we introduce a multimodal mixed self-attention mechanism to enable cross-modal interactions, further enhancing information fusion. Extensive experiments on five benchmark cancer datasets demonstrate the superiority of our approach over existing methods. The code is available at https://github.com/HongLiuuuuu/APL.
comment: 10 pages, 3 figures
☆ Simulating the Real World: A Unified Survey of Multimodal Generative Models
Understanding and replicating the real world is a critical challenge in Artificial General Intelligence (AGI) research. To achieve this, many existing approaches, such as world models, aim to capture the fundamental principles governing the physical world, enabling more accurate simulations and meaningful interactions. However, current methods often treat different modalities, including 2D (images), videos, 3D, and 4D representations, as independent domains, overlooking their interdependencies. Additionally, these methods typically focus on isolated dimensions of reality without systematically integrating their connections. In this survey, we present a unified survey for multimodal generative models that investigate the progression of data dimensionality in real-world simulation. Specifically, this survey starts from 2D generation (appearance), then moves to video (appearance+dynamics) and 3D generation (appearance+geometry), and finally culminates in 4D generation that integrate all dimensions. To the best of our knowledge, this is the first attempt to systematically unify the study of 2D, video, 3D and 4D generation within a single framework. To guide future research, we provide a comprehensive review of datasets, evaluation metrics and future directions, and fostering insights for newcomers. This survey serves as a bridge to advance the study of multimodal generative models and real-world simulation within a unified framework.
comment: Repository for the related papers at https://github.com/ALEEEHU/World-Simulator
☆ Enhancing SAM with Efficient Prompting and Preference Optimization for Semi-supervised Medical Image Segmentation CVPR 2025
Foundational models such as the Segment Anything Model (SAM) are gaining traction in medical imaging segmentation, supporting multiple downstream tasks. However, such models are supervised in nature, still relying on large annotated datasets or prompts supplied by experts. Conventional techniques such as active learning to alleviate such limitations are limited in scope and still necessitate continuous human involvement and complex domain knowledge for label refinement or establishing reward ground truth. To address these challenges, we propose an enhanced Segment Anything Model (SAM) framework that utilizes annotation-efficient prompts generated in a fully unsupervised fashion, while still capturing essential semantic, location, and shape information through contrastive language-image pretraining and visual question answering. We adopt the direct preference optimization technique to design an optimal policy that enables the model to generate high-fidelity segmentations with simple ratings or rankings provided by a virtual annotator simulating the human annotation process. State-of-the-art performance of our framework in tasks such as lung segmentation, breast tumor segmentation, and organ segmentation across various modalities, including X-ray, ultrasound, and abdominal CT, justifies its effectiveness in low-annotation data scenarios.
comment: Accepted to CVPR 2025
☆ 3HANDS Dataset: Learning from Humans for Generating Naturalistic Handovers with Supernumerary Robotic Limbs
Supernumerary robotic limbs (SRLs) are robotic structures integrated closely with the user's body, which augment human physical capabilities and necessitate seamless, naturalistic human-machine interaction. For effective assistance in physical tasks, enabling SRLs to hand over objects to humans is crucial. Yet, designing heuristic-based policies for robots is time-consuming, difficult to generalize across tasks, and results in less human-like motion. When trained with proper datasets, generative models are powerful alternatives for creating naturalistic handover motions. We introduce 3HANDS, a novel dataset of object handover interactions between a participant performing a daily activity and another participant enacting a hip-mounted SRL in a naturalistic manner. 3HANDS captures the unique characteristics of SRL interactions: operating in intimate personal space with asymmetric object origins, implicit motion synchronization, and the user's engagement in a primary task during the handover. To demonstrate the effectiveness of our dataset, we present three models: one that generates naturalistic handover trajectories, another that determines the appropriate handover endpoints, and a third that predicts the moment to initiate a handover. In a user study (N=10), we compare the handover interaction performed with our method compared to a baseline. The findings show that our method was perceived as significantly more natural, less physically demanding, and more comfortable.
comment: CHI '25
☆ PathoPainter: Augmenting Histopathology Segmentation via Tumor-aware Inpainting
Tumor segmentation plays a critical role in histopathology, but it requires costly, fine-grained image-mask pairs annotated by pathologists. Thus, synthesizing histopathology data to expand the dataset is highly desirable. Previous works suffer from inaccuracies and limited diversity in image-mask pairs, both of which affect training segmentation, particularly in small-scale datasets and the inherently complex nature of histopathology images. To address this challenge, we propose PathoPainter, which reformulates image-mask pair generation as a tumor inpainting task. Specifically, our approach preserves the background while inpainting the tumor region, ensuring precise alignment between the generated image and its corresponding mask. To enhance dataset diversity while maintaining biological plausibility, we incorporate a sampling mechanism that conditions tumor inpainting on regional embeddings from a different image. Additionally, we introduce a filtering strategy to exclude uncertain synthetic regions, further improving the quality of the generated data. Our comprehensive evaluation spans multiple datasets featuring diverse tumor types and various training data scales. As a result, segmentation improved significantly with our synthetic data, surpassing existing segmentation data synthesis approaches, e.g., 75.69% -> 77.69% on CAMELYON16. The code is available at https://github.com/HongLiuuuuu/PathoPainter.
comment: 10 pages, 3 figures
☆ The Best of Both Worlds: Integrating Language Models and Diffusion Models for Video Generation
Recent advancements in text-to-video (T2V) generation have been driven by two competing paradigms: autoregressive language models and diffusion models. However, each paradigm has intrinsic limitations: language models struggle with visual quality and error accumulation, while diffusion models lack semantic understanding and causal modeling. In this work, we propose LanDiff, a hybrid framework that synergizes the strengths of both paradigms through coarse-to-fine generation. Our architecture introduces three key innovations: (1) a semantic tokenizer that compresses 3D visual features into compact 1D discrete representations through efficient semantic compression, achieving a $\sim$14,000$\times$ compression ratio; (2) a language model that generates semantic tokens with high-level semantic relationships; (3) a streaming diffusion model that refines coarse semantics into high-fidelity videos. Experiments show that LanDiff, a 5B model, achieves a score of 85.43 on the VBench T2V benchmark, surpassing the state-of-the-art open-source models Hunyuan Video (13B) and other commercial models such as Sora, Keling, and Hailuo. Furthermore, our model also achieves state-of-the-art performance in long video generation, surpassing other open-source models in this field. Our demo can be viewed at https://landiff.github.io/.
☆ A Benchmark for Multi-Lingual Vision-Language Learning in Remote Sensing Image Captioning
Remote Sensing Image Captioning (RSIC) is a cross-modal field bridging vision and language, aimed at automatically generating natural language descriptions of features and scenes in remote sensing imagery. Despite significant advances in developing sophisticated methods and large-scale datasets for training vision-language models (VLMs), two critical challenges persist: the scarcity of non-English descriptive datasets and the lack of multilingual capability evaluation for models. These limitations fundamentally impede the progress and practical deployment of RSIC, particularly in the era of large VLMs. To address these challenges, this paper presents several significant contributions to the field. First, we introduce and analyze BRSIC (Bilingual Remote Sensing Image Captioning), a comprehensive bilingual dataset that enriches three established English RSIC datasets with Chinese descriptions, encompassing 13,634 images paired with 68,170 bilingual captions. Building upon this foundation, we develop a systematic evaluation framework that addresses the prevalent inconsistency in evaluation protocols, enabling rigorous assessment of model performance through standardized retraining procedures on BRSIC. Furthermore, we present an extensive empirical study of eight state-of-the-art large vision-language models (LVLMs), examining their capabilities across multiple paradigms including zero-shot inference, supervised fine-tuning, and multi-lingual training. This comprehensive evaluation provides crucial insights into the strengths and limitations of current LVLMs in handling multilingual remote sensing tasks. Additionally, our cross-dataset transfer experiments reveal interesting findings. The code and data will be available at https://github.com/mrazhou/BRSIC.
☆ Omnidirectional Multi-Object Tracking CVPR 2025
Panoramic imagery, with its 360{\deg} field of view, offers comprehensive information to support Multi-Object Tracking (MOT) in capturing spatial and temporal relationships of surrounding objects. However, most MOT algorithms are tailored for pinhole images with limited views, impairing their effectiveness in panoramic settings. Additionally, panoramic image distortions, such as resolution loss, geometric deformation, and uneven lighting, hinder direct adaptation of existing MOT methods, leading to significant performance degradation. To address these challenges, we propose OmniTrack, an omnidirectional MOT framework that incorporates Tracklet Management to introduce temporal cues, FlexiTrack Instances for object localization and association, and the CircularStatE Module to alleviate image and geometric distortions. This integration enables tracking in large field-of-view scenarios, even under rapid sensor motion. To mitigate the lack of panoramic MOT datasets, we introduce the QuadTrack dataset--a comprehensive panoramic dataset collected by a quadruped robot, featuring diverse challenges such as wide fields of view, intense motion, and complex environments. Extensive experiments on the public JRDB dataset and the newly introduced QuadTrack benchmark demonstrate the state-of-the-art performance of the proposed framework. OmniTrack achieves a HOTA score of 26.92% on JRDB, representing an improvement of 3.43%, and further achieves 23.45% on QuadTrack, surpassing the baseline by 6.81%. The dataset and code will be made publicly available at https://github.com/xifen523/OmniTrack.
comment: Accepted to CVPR 2025. The dataset and code will be made publicly available at https://github.com/xifen523/OmniTrack
☆ ViT-VS: On the Applicability of Pretrained Vision Transformer Features for Generalizable Visual Servoing
Visual servoing enables robots to precisely position their end-effector relative to a target object. While classical methods rely on hand-crafted features and thus are universally applicable without task-specific training, they often struggle with occlusions and environmental variations, whereas learning-based approaches improve robustness but typically require extensive training. We present a visual servoing approach that leverages pretrained vision transformers for semantic feature extraction, combining the advantages of both paradigms while also being able to generalize beyond the provided sample. Our approach achieves full convergence in unperturbed scenarios and surpasses classical image-based visual servoing by up to 31.2\% relative improvement in perturbed scenarios. Even the convergence rates of learning-based methods are matched despite requiring no task- or object-specific training. Real-world evaluations confirm robust performance in end-effector positioning, industrial box manipulation, and grasping of unseen objects using only a reference from the same category. Our code and simulation environment are available at: https://alessandroscherl.github.io/ViT-VS/
☆ In-Context Reverse Classification Accuracy: Efficient Estimation of Segmentation Quality without Ground-Truth
Assessing the quality of automatic image segmentation is crucial in clinical practice, but often very challenging due to the limited availability of ground truth annotations. In this paper, we introduce In-Context Reverse Classification Accuracy (In-Context RCA), a novel framework for automatically estimating segmentation quality in the absence of ground-truth annotations. By leveraging recent in-context learning segmentation models and incorporating retrieval-augmentation techniques to select the most relevant reference images, our approach enables efficient quality estimation with minimal reference data. Validated across diverse medical imaging modalities, our method demonstrates robust performance and computational efficiency, offering a promising solution for automated quality control in clinical workflows, where fast and reliable segmentation assessment is essential. The code is available at https://github.com/mcosarinsky/In-Context-RCA.
☆ A Novel Solution for Drone Photogrammetry with Low-overlap Aerial Images using Monocular Depth Estimation
Low-overlap aerial imagery poses significant challenges to traditional photogrammetric methods, which rely heavily on high image overlap to produce accurate and complete mapping products. In this study, we propose a novel workflow based on monocular depth estimation to address the limitations of conventional techniques. Our method leverages tie points obtained from aerial triangulation to establish a relationship between monocular depth and metric depth, thus transforming the original depth map into a metric depth map, enabling the generation of dense depth information and the comprehensive reconstruction of the scene. For the experiments, a high-overlap drone dataset containing 296 images is processed using Metashape to generate depth maps and DSMs as ground truth. Subsequently, we create a low-overlap dataset by selecting 20 images for experimental evaluation. Results demonstrate that while the recovered depth maps and resulting DSMs achieve meter-level accuracy, they provide significantly better completeness compared to traditional methods, particularly in regions covered by single images. This study showcases the potential of monocular depth estimation in low-overlap aerial photogrammetry.
☆ AnyAnomaly: Zero-Shot Customizable Video Anomaly Detection with LVLM
Video anomaly detection (VAD) is crucial for video analysis and surveillance in computer vision. However, existing VAD models rely on learned normal patterns, which makes them difficult to apply to diverse environments. Consequently, users should retrain models or develop separate AI models for new environments, which requires expertise in machine learning, high-performance hardware, and extensive data collection, limiting the practical usability of VAD. To address these challenges, this study proposes customizable video anomaly detection (C-VAD) technique and the AnyAnomaly model. C-VAD considers user-defined text as an abnormal event and detects frames containing a specified event in a video. We effectively implemented AnyAnomaly using a context-aware visual question answering without fine-tuning the large vision language model. To validate the effectiveness of the proposed model, we constructed C-VAD datasets and demonstrated the superiority of AnyAnomaly. Furthermore, our approach showed competitive performance on VAD benchmark datasets, achieving state-of-the-art results on the UBnormal dataset and outperforming other methods in generalization across all datasets. Our code is available online at github.com/SkiddieAhn/Paper-AnyAnomaly.
☆ IMFine: 3D Inpainting via Geometry-guided Multi-view Refinement CVPR 2025
Current 3D inpainting and object removal methods are largely limited to front-facing scenes, facing substantial challenges when applied to diverse, "unconstrained" scenes where the camera orientation and trajectory are unrestricted. To bridge this gap, we introduce a novel approach that produces inpainted 3D scenes with consistent visual quality and coherent underlying geometry across both front-facing and unconstrained scenes. Specifically, we propose a robust 3D inpainting pipeline that incorporates geometric priors and a multi-view refinement network trained via test-time adaptation, building on a pre-trained image inpainting model. Additionally, we develop a novel inpainting mask detection technique to derive targeted inpainting masks from object masks, boosting the performance in handling unconstrained scenes. To validate the efficacy of our approach, we create a challenging and diverse benchmark that spans a wide range of scenes. Comprehensive experiments demonstrate that our proposed method substantially outperforms existing state-of-the-art approaches.
comment: Accepted at CVPR 2025, \href{https://xinxinzuo2353.github.io/imfine/}{Project Page}
☆ ReynoldsFlow: Exquisite Flow Estimation via Reynolds Transport Theorem
Optical flow is a fundamental technique for motion estimation, widely applied in video stabilization, interpolation, and object tracking. Recent advancements in artificial intelligence (AI) have enabled deep learning models to leverage optical flow as an important feature for motion analysis. However, traditional optical flow methods rely on restrictive assumptions, such as brightness constancy and slow motion constraints, limiting their effectiveness in complex scenes. Deep learning-based approaches require extensive training on large domain-specific datasets, making them computationally demanding. Furthermore, optical flow is typically visualized in the HSV color space, which introduces nonlinear distortions when converted to RGB and is highly sensitive to noise, degrading motion representation accuracy. These limitations inherently constrain the performance of downstream models, potentially hindering object tracking and motion analysis tasks. To address these challenges, we propose Reynolds flow, a novel training-free flow estimation inspired by the Reynolds transport theorem, offering a principled approach to modeling complex motion dynamics. Beyond the conventional HSV-based visualization, denoted ReynoldsFlow, we introduce an alternative representation, ReynoldsFlow+, designed to improve flow visualization. We evaluate ReynoldsFlow and ReynoldsFlow+ across three video-based benchmarks: tiny object detection on UAVDB, infrared object detection on Anti-UAV, and pose estimation on GolfDB. Experimental results demonstrate that networks trained with ReynoldsFlow+ achieve state-of-the-art (SOTA) performance, exhibiting improved robustness and efficiency across all tasks.
comment: 10 pages, 3 figures, 3 tables
☆ Spatial regularisation for improved accuracy and interpretability in keypoint-based registration
Unsupervised registration strategies bypass requirements in ground truth transforms or segmentations by optimising similarity metrics between fixed and moved volumes. Among these methods, a recent subclass of approaches based on unsupervised keypoint detection stand out as very promising for interpretability. Specifically, these methods train a network to predict feature maps for fixed and moving images, from which explainable centres of mass are computed to obtain point clouds, that are then aligned in closed-form. However, the features returned by the network often yield spatially diffuse patterns that are hard to interpret, thus undermining the purpose of keypoint-based registration. Here, we propose a three-fold loss to regularise the spatial distribution of the features. First, we use the KL divergence to model features as point spread functions that we interpret as probabilistic keypoints. Then, we sharpen the spatial distributions of these features to increase the precision of the detected landmarks. Finally, we introduce a new repulsive loss across keypoints to encourage spatial diversity. Overall, our loss considerably improves the interpretability of the features, which now correspond to precise and anatomically meaningful landmarks. We demonstrate our three-fold loss in foetal rigid motion tracking and brain MRI affine registration tasks, where it not only outperforms state-of-the-art unsupervised strategies, but also bridges the gap with state-of-the-art supervised methods. Our code is available at https://github.com/BenBillot/spatial_regularisation.
comment: under review
☆ Learning Object Placement Programs for Indoor Scene Synthesis with Iterative Self Training
Data driven and autoregressive indoor scene synthesis systems generate indoor scenes automatically by suggesting and then placing objects one at a time. Empirical observations show that current systems tend to produce incomplete next object location distributions. We introduce a system which addresses this problem. We design a Domain Specific Language (DSL) that specifies functional constraints. Programs from our language take as input a partial scene and object to place. Upon execution they predict possible object placements. We design a generative model which writes these programs automatically. Available 3D scene datasets do not contain programs to train on, so we build upon previous work in unsupervised program induction to introduce a new program bootstrapping algorithm. In order to quantify our empirical observations we introduce a new evaluation procedure which captures how well a system models per-object location distributions. We ask human annotators to label all the possible places an object can go in a scene and show that our system produces per-object location distributions more consistent with human annotators. Our system also generates indoor scenes of comparable quality to previous systems and while previous systems degrade in performance when training data is sparse, our system does not degrade to the same degree.
comment: 21 pages, 20 figures Subjects: Graphics (cs.GR), Computer Vision and Pattern Recognition (cs.CV), Machine Learning (cs.LG)
☆ Semantic Alignment of Unimodal Medical Text and Vision Representations
General-purpose AI models, particularly those designed for text and vision, demonstrate impressive versatility across a wide range of deep-learning tasks. However, they often underperform in specialised domains like medical imaging, where domain-specific solutions or alternative knowledge transfer approaches are typically required. Recent studies have noted that general-purpose models can exhibit similar latent spaces when processing semantically related data, although this alignment does not occur naturally. Building on this insight, it has been shown that applying a simple transformation - at most affine - estimated from a subset of semantically corresponding samples, known as anchors, enables model stitching across diverse training paradigms, architectures, and modalities. In this paper, we explore how semantic alignment - estimating transformations between anchors - can bridge general-purpose AI with specialised medical knowledge. Using multiple public chest X-ray datasets, we demonstrate that model stitching across model architectures allows general models to integrate domain-specific knowledge without additional training, leading to improved performance on medical tasks. Furthermore, we introduce a novel zero-shot classification approach for unimodal vision encoders that leverages semantic alignment across modalities. Our results show that our method not only outperforms general multimodal models but also approaches the performance levels of fully trained, medical-specific multimodal solutions
☆ ForestLPR: LiDAR Place Recognition in Forests Attentioning Multiple BEV Density Images CVPR2025
Place recognition is essential to maintain global consistency in large-scale localization systems. While research in urban environments has progressed significantly using LiDARs or cameras, applications in natural forest-like environments remain largely under-explored. Furthermore, forests present particular challenges due to high self-similarity and substantial variations in vegetation growth over time. In this work, we propose a robust LiDAR-based place recognition method for natural forests, ForestLPR. We hypothesize that a set of cross-sectional images of the forest's geometry at different heights contains the information needed to recognize revisiting a place. The cross-sectional images are represented by \ac{bev} density images of horizontal slices of the point cloud at different heights. Our approach utilizes a visual transformer as the shared backbone to produce sets of local descriptors and introduces a multi-BEV interaction module to attend to information at different heights adaptively. It is followed by an aggregation layer that produces a rotation-invariant place descriptor. We evaluated the efficacy of our method extensively on real-world data from public benchmarks as well as robotic datasets and compared it against the state-of-the-art (SOTA) methods. The results indicate that ForestLPR has consistently good performance on all evaluations and achieves an average increase of 7.38\% and 9.11\% on Recall@1 over the closest competitor on intra-sequence loop closure detection and inter-sequence re-localization, respectively, validating our hypothesis
comment: accepted by CVPR2025
☆ Gate-Shift-Pose: Enhancing Action Recognition in Sports with Skeleton Information
This paper introduces Gate-Shift-Pose, an enhanced version of Gate-Shift-Fuse networks, designed for athlete fall classification in figure skating by integrating skeleton pose data alongside RGB frames. We evaluate two fusion strategies: early-fusion, which combines RGB frames with Gaussian heatmaps of pose keypoints at the input stage, and late-fusion, which employs a multi-stream architecture with attention mechanisms to combine RGB and pose features. Experiments on the FR-FS dataset demonstrate that Gate-Shift-Pose significantly outperforms the RGB-only baseline, improving accuracy by up to 40% with ResNet18 and 20% with ResNet50. Early-fusion achieves the highest accuracy (98.08%) with ResNet50, leveraging the model's capacity for effective multimodal integration, while late-fusion is better suited for lighter backbones like ResNet18. These results highlight the potential of multimodal architectures for sports action recognition and the critical role of skeleton pose information in capturing complex motion patterns.
☆ Question-Aware Gaussian Experts for Audio-Visual Question Answering CVPR 2025
Audio-Visual Question Answering (AVQA) requires not only question-based multimodal reasoning but also precise temporal grounding to capture subtle dynamics for accurate prediction. However, existing methods mainly use question information implicitly, limiting focus on question-specific details. Furthermore, most studies rely on uniform frame sampling, which can miss key question-relevant frames. Although recent Top-K frame selection methods aim to address this, their discrete nature still overlooks fine-grained temporal details. This paper proposes \textbf{QA-TIGER}, a novel framework that explicitly incorporates question information and models continuous temporal dynamics. Our key idea is to use Gaussian-based modeling to adaptively focus on both consecutive and non-consecutive frames based on the question, while explicitly injecting question information and applying progressive refinement. We leverage a Mixture of Experts (MoE) to flexibly implement multiple Gaussian models, activating temporal experts specifically tailored to the question. Extensive experiments on multiple AVQA benchmarks show that QA-TIGER consistently achieves state-of-the-art performance. Code is available at https://github.com/AIM-SKKU/QA-TIGER
comment: CVPR 2025. Project page at https://aim-skku.github.io/QA-TIGER/
☆ TPC: Cross-Temporal Prediction Connection for Vision-Language Model Hallucination Reduction
Vision-language models (VLMs) have achieved remarkable advancements, capitalizing on the impressive capabilities of large language models (LLMs) across diverse tasks. Despite this, a critical challenge known as hallucination occurs when models overconfidently describe objects or attributes absent from the image, a problem exacerbated by the tendency of VLMs to rely on linguistic priors. This limitation reduces model reliability in high-stakes applications. In this work, we have observed the characteristic of logits' continuity consistency enhancement and introduced a straightforward and efficient method, Cross-Temporal Prediction Connection (TPC), designed to enhance the semantic consistency of logits by connecting them temporally across timesteps. TPC amplifies information flow and improves coherence, effectively reducing hallucination. Extensive experiments show that TPC surpasses existing representatives, delivering superior performance in both accuracy and efficiency while maintaining robustness in open-ended text generation tasks.
☆ A lightweight model FDM-YOLO for small target improvement based on YOLOv8
Small targets are particularly difficult to detect due to their low pixel count, complex backgrounds, and varying shooting angles, which make it hard for models to extract effective features. While some large-scale models offer high accuracy, their long inference times make them unsuitable for real-time deployment on edge devices. On the other hand, models designed for low computational power often suffer from poor detection accuracy. This paper focuses on small target detection and explores methods for object detection under low computational constraints. Building on the YOLOv8 model, we propose a new network architecture called FDM-YOLO. Our research includes the following key contributions: We introduce FDM-YOLO by analyzing the output of the YOLOv8 detection head. We add a highresolution layer and remove the large target detection layer to better handle small targets. Based on PConv, we propose a lightweight network structure called Fast-C2f, which is integrated into the PAN module of the model. To mitigate the accuracy loss caused by model lightweighting, we employ dynamic upsampling (Dysample) and a lightweight EMA attention mechanism.The FDM-YOLO model was validated on the Visdrone dataset, achieving a 38% reduction in parameter count and improving the Map0.5 score from 38.4% to 42.5%, all while maintaining nearly the same inference speed. This demonstrates the effectiveness of our approach in balancing accuracy and efficiency for edge device deployment.
☆ ToFu: Visual Tokens Reduction via Fusion for Multi-modal, Multi-patch, Multi-image Task
Large Multimodal Models (LMMs) are powerful tools that are capable of reasoning and understanding multimodal information beyond text and language. Despite their entrenched impact, the development of LMMs is hindered by the higher computational requirements compared to their unimodal counterparts. One of the main causes of this is the large amount of tokens needed to encode the visual input, which is especially evident for multi-image multimodal tasks. Recent approaches to reduce visual tokens depend on the visual encoder architecture, require fine-tuning the LLM to maintain the performance, and only consider single-image scenarios. To address these limitations, we propose ToFu, a visual encoder-agnostic, training-free Token Fusion strategy that combines redundant visual tokens of LMMs for high-resolution, multi-image, tasks. The core intuition behind our method is straightforward yet effective: preserve distinctive tokens while combining similar ones. We achieve this by sequentially examining visual tokens and deciding whether to merge them with others or keep them as separate entities. We validate our approach on the well-established LLaVA-Interleave Bench, which covers challenging multi-image tasks. In addition, we push to the extreme our method by testing it on a newly-created benchmark, ComPairs, focused on multi-image comparisons where a larger amount of images and visual tokens are inputted to the LMMs. Our extensive analysis, considering several LMM architectures, demonstrates the benefits of our approach both in terms of efficiency and performance gain.
☆ EvidMTL: Evidential Multi-Task Learning for Uncertainty-Aware Semantic Surface Mapping from Monocular RGB Images IROS 2025
For scene understanding in unstructured environments, an accurate and uncertainty-aware metric-semantic mapping is required to enable informed action selection by autonomous systems.Existing mapping methods often suffer from overconfident semantic predictions, and sparse and noisy depth sensing, leading to inconsistent map representations. In this paper, we therefore introduce EvidMTL, a multi-task learning framework that uses evidential heads for depth estimation and semantic segmentation, enabling uncertainty-aware inference from monocular RGB images. To enable uncertainty-calibrated evidential multi-task learning, we propose a novel evidential depth loss function that jointly optimizes the belief strength of the depth prediction in conjunction with evidential segmentation loss. Building on this, we present EvidKimera, an uncertainty-aware semantic surface mapping framework, which uses evidential depth and semantics prediction for improved 3D metric-semantic consistency. We train and evaluate EvidMTL on the NYUDepthV2 and assess its zero-shot performance on ScanNetV2, demonstrating superior uncertainty estimation compared to conventional approaches while maintaining comparable depth estimation and semantic segmentation. In zero-shot mapping tests on ScanNetV2, EvidKimera outperforms Kimera in semantic surface mapping accuracy and consistency, highlighting the benefits of uncertainty-aware mapping and underscoring its potential for real-world robotic applications.
comment: Submitted to IROS 2025 Conference
☆ PointsToWood: A deep learning framework for complete canopy leaf-wood segmentation of TLS data across diverse European forests
Point clouds from Terrestrial Laser Scanning (TLS) are an increasingly popular source of data for studying plant structure and function but typically require extensive manual processing to extract ecologically important information. One key task is the accurate semantic segmentation of different plant material within point clouds, particularly wood and leaves, which is required to understand plant productivity, architecture and physiology. Existing automated semantic segmentation methods are primarily developed for single ecosystem types, and whilst they show good accuracy for biomass assessment from the trunk and large branches, often perform less well within the crown. In this study, we demonstrate a new framework that uses a deep learning architecture newly developed from PointNet and pointNEXT for processing 3D point clouds to provide a reliable semantic segmentation of wood and leaf in TLS point clouds from the tree base to branch tips, trained on data from diverse mature European forests. Our model uses meticulously labelled data combined with voxel-based sampling, neighbourhood rescaling, and a novel gated reflectance integration module embedded throughout the feature extraction layers. We evaluate its performance across open datasets from boreal, temperate, Mediterranean and tropical regions, encompassing diverse ecosystem types and sensor characteristics. Our results show consistent outperformance against the most widely used PointNet based approach for leaf/wood segmentation on our high-density TLS dataset collected across diverse mixed forest plots across all major biomes in Europe. We also find consistently strong performance tested on others open data from China, Eastern Cameroon, Germany and Finland, collected using both time-of-flight and phase-shift sensors, showcasing the transferability of our model to a wide range of ecosystems and sensors.
☆ Learning Transformer-based World Models with Contrastive Predictive Coding
The DreamerV3 algorithm recently obtained remarkable performance across diverse environment domains by learning an accurate world model based on Recurrent Neural Networks (RNNs). Following the success of model-based reinforcement learning algorithms and the rapid adoption of the Transformer architecture for its superior training efficiency and favorable scaling properties, recent works such as STORM have proposed replacing RNN-based world models with Transformer-based world models using masked self-attention. However, despite the improved training efficiency of these methods, their impact on performance remains limited compared to the Dreamer algorithm, struggling to learn competitive Transformer-based world models. In this work, we show that the next state prediction objective adopted in previous approaches is insufficient to fully exploit the representation capabilities of Transformers. We propose to extend world model predictions to longer time horizons by introducing TWISTER (Transformer-based World model wIth contraSTivE Representations), a world model using action-conditioned Contrastive Predictive Coding to learn high-level temporal feature representations and improve the agent performance. TWISTER achieves a human-normalized mean score of 162% on the Atari 100k benchmark, setting a new record among state-of-the-art methods that do not employ look-ahead search.
☆ Scale-Invariant Adversarial Attack against Arbitrary-scale Super-resolution
The advent of local continuous image function (LIIF) has garnered significant attention for arbitrary-scale super-resolution (SR) techniques. However, while the vulnerabilities of fixed-scale SR have been assessed, the robustness of continuous representation-based arbitrary-scale SR against adversarial attacks remains an area warranting further exploration. The elaborately designed adversarial attacks for fixed-scale SR are scale-dependent, which will cause time-consuming and memory-consuming problems when applied to arbitrary-scale SR. To address this concern, we propose a simple yet effective ``scale-invariant'' SR adversarial attack method with good transferability, termed SIAGT. Specifically, we propose to construct resource-saving attacks by exploiting finite discrete points of continuous representation. In addition, we formulate a coordinate-dependent loss to enhance the cross-model transferability of the attack. The attack can significantly deteriorate the SR images while introducing imperceptible distortion to the targeted low-resolution (LR) images. Experiments carried out on three popular LIIF-based SR approaches and four classical SR datasets show remarkable attack performance and transferability of SIAGT.
comment: 15 pages, accepted by TIFS 2025
☆ MIDAS: Modeling Ground-Truth Distributions with Dark Knowledge for Domain Generalized Stereo Matching
Despite the significant advances in domain generalized stereo matching, existing methods still exhibit domain-specific preferences when transferring from synthetic to real domains, hindering their practical applications in complex and diverse scenarios. The probability distributions predicted by the stereo network naturally encode rich similarity and uncertainty information. Inspired by this observation, we propose to extract these two types of dark knowledge from the pre-trained network to model intuitive multi-modal ground-truth distributions for both edge and non-edge regions. To mitigate the inherent domain preferences of a single network, we adopt network ensemble and further distinguish between objective and biased knowledge in the Laplace parameter space. Finally, the objective knowledge and the original disparity labels are jointly modeled as a mixture of Laplacians to provide fine-grained supervision for the stereo network training. Extensive experiments demonstrate that: 1) Our method is generic and effectively improves the generalization of existing networks. 2) PCWNet with our method achieves the state-of-the-art generalization performance on both KITTI 2015 and 2012 datasets. 3) Our method outperforms existing methods in comprehensive ranking across four popular real-world datasets.
☆ ObjMST: An Object-Focused Multimodal Style Transfer Framework
We propose ObjMST, an object-focused multimodal style transfer framework that provides separate style supervision for salient objects and surrounding elements while addressing alignment issues in multimodal representation learning. Existing image-text multimodal style transfer methods face the following challenges: (1) generating non-aligned and inconsistent multimodal style representations; and (2) content mismatch, where identical style patterns are applied to both salient objects and their surrounding elements. Our approach mitigates these issues by: (1) introducing a Style-Specific Masked Directional CLIP Loss, which ensures consistent and aligned style representations for both salient objects and their surroundings; and (2) incorporating a salient-to-key mapping mechanism for stylizing salient objects, followed by image harmonization to seamlessly blend the stylized objects with their environment. We validate the effectiveness of ObjMST through experiments, using both quantitative metrics and qualitative visual evaluations of the stylized outputs. Our code is available at: https://github.com/chandagrover/ObjMST.
comment: 8 pages, 8 Figures, 3 Tables
☆ PLMP -- Point-Line Minimal Problems for Projective SfM
We completely classify all minimal problems for Structure-from-Motion (SfM) where arrangements of points and lines are fully observed by multiple uncalibrated pinhole cameras. We find 291 minimal problems, 73 of which have unique solutions and can thus be solved linearly. Two of the linear problems allow an arbitrary number of views, while all other minimal problems have at most 9 cameras. All minimal problems have at most 7 points and at most 12 lines. We compute the number of solutions of each minimal problem, as this gives a measurement of the problem's intrinsic difficulty, and find that these number are relatively low (e.g., when comparing with minimal problems for calibrated cameras). Finally, by exploring stabilizer subgroups of subarrangements, we develop a geometric and systematic way to 1) factorize minimal problems into smaller problems, 2) identify minimal problems in underconstrained problems, and 3) formally prove non-minimality.
☆ LEDiT: Your Length-Extrapolatable Diffusion Transformer without Positional Encoding
Diffusion transformers(DiTs) struggle to generate images at resolutions higher than their training resolutions. The primary obstacle is that the explicit positional encodings(PE), such as RoPE, need extrapolation which degrades performance when the inference resolution differs from training. In this paper, we propose a Length-Extrapolatable Diffusion Transformer(LEDiT), a simple yet powerful architecture to overcome this limitation. LEDiT needs no explicit PEs, thereby avoiding extrapolation. The key innovations of LEDiT are introducing causal attention to implicitly impart global positional information to tokens, while enhancing locality to precisely distinguish adjacent tokens. Experiments on 256x256 and 512x512 ImageNet show that LEDiT can scale the inference resolution to 512x512 and 1024x1024, respectively, while achieving better image quality compared to current state-of-the-art length extrapolation methods(NTK-aware, YaRN). Moreover, LEDiT achieves strong extrapolation performance with just 100K steps of fine-tuning on a pretrained DiT, demonstrating its potential for integration into existing text-to-image DiTs.
☆ GaussianVideo: Efficient Video Representation and Compression by Gaussian Splatting
Implicit Neural Representation for Videos (NeRV) has introduced a novel paradigm for video representation and compression, outperforming traditional codecs. As model size grows, however, slow encoding and decoding speed and high memory consumption hinder its application in practice. To address these limitations, we propose a new video representation and compression method based on 2D Gaussian Splatting to efficiently handle video data. Our proposed deformable 2D Gaussian Splatting dynamically adapts the transformation of 2D Gaussians at each frame, significantly reducing memory cost. Equipped with a multi-plane-based spatiotemporal encoder and a lightweight decoder, it predicts changes in color, coordinates, and shape of initialized Gaussians, given the time step. By leveraging temporal gradients, our model effectively captures temporal redundancy at negligible cost, significantly enhancing video representation efficiency. Our method reduces GPU memory usage by up to 78.4%, and significantly expedites video processing, achieving 5.5x faster training and 12.5x faster decoding compared to the state-of-the-art NeRV methods.
☆ GBT-SAM: A Parameter-Efficient Depth-Aware Model for Generalizable Brain tumour Segmentation on mp-MRI
Gliomas are brain tumours that stand out for their highly lethal and aggressive nature, which demands a precise approach in their diagnosis. Medical image segmentation plays a crucial role in the evaluation and follow-up of these tumours, allowing specialists to analyse their morphology. However, existing methods for automatic glioma segmentation often lack generalization capability across other brain tumour domains, require extensive computational resources, or fail to fully utilize the multi-parametric MRI (mp-MRI) data used to delineate them. In this work, we introduce GBT-SAM, a novel Generalizable Brain Tumour (GBT) framework that extends the Segment Anything Model (SAM) to brain tumour segmentation tasks. Our method employs a two-step training protocol: first, fine-tuning the patch embedding layer to process the entire mp-MRI modalities, and second, incorporating parameter-efficient LoRA blocks and a Depth-Condition block into the Vision Transformer (ViT) to capture inter-slice correlations. GBT-SAM achieves state-of-the-art performance on the Adult Glioma dataset (Dice Score of $93.54$) while demonstrating robust generalization across Meningioma, Pediatric Glioma, and Sub-Saharan Glioma datasets. Furthermore, GBT-SAM uses less than 6.5M trainable parameters, thus offering an efficient solution for brain tumour segmentation. \\ Our code and models are available at https://github.com/vpulab/med-sam-brain .
☆ A Modular Pipeline for 3D Object Tracking Using RGB Cameras
Object tracking is a key challenge of computer vision with various applications that all require different architectures. Most tracking systems have limitations such as constraining all movement to a 2D plane and they often track only one object. In this paper, we present a new modular pipeline that calculates 3D trajectories of multiple objects. It is adaptable to various settings where multiple time-synced and stationary cameras record moving objects, using off the shelf webcams. Our pipeline was tested on the Table Setting Dataset, where participants are recorded with various sensors as they set a table with tableware objects. We need to track these manipulated objects, using 6 rgb webcams. Challenges include: Detecting small objects in 9.874.699 camera frames, determining camera poses, discriminating between nearby and overlapping objects, temporary occlusions, and finally calculating a 3D trajectory using the right subset of an average of 11.12.456 pixel coordinates per 3-minute trial. We implement a robust pipeline that results in accurate trajectories with covariance of x,y,z-position as a confidence metric. It deals dynamically with appearing and disappearing objects, instantiating new Extended Kalman Filters. It scales to hundreds of table-setting trials with very little human annotation input, even with the camera poses of each trial unknown. The code is available at https://github.com/LarsBredereke/object_tracking
comment: 9 pages, 11 figures, original paper not to be published anywhere else
☆ S2Gaussian: Sparse-View Super-Resolution 3D Gaussian Splatting CVPR 2025
In this paper, we aim ambitiously for a realistic yet challenging problem, namely, how to reconstruct high-quality 3D scenes from sparse low-resolution views that simultaneously suffer from deficient perspectives and clarity. Whereas existing methods only deal with either sparse views or low-resolution observations, they fail to handle such hybrid and complicated scenarios. To this end, we propose a novel Sparse-view Super-resolution 3D Gaussian Splatting framework, dubbed S2Gaussian, that can reconstruct structure-accurate and detail-faithful 3D scenes with only sparse and low-resolution views. The S2Gaussian operates in a two-stage fashion. In the first stage, we initially optimize a low-resolution Gaussian representation with depth regularization and densify it to initialize the high-resolution Gaussians through a tailored Gaussian Shuffle Split operation. In the second stage, we refine the high-resolution Gaussians with the super-resolved images generated from both original sparse views and pseudo-views rendered by the low-resolution Gaussians. In which a customized blur-free inconsistency modeling scheme and a 3D robust optimization strategy are elaborately designed to mitigate multi-view inconsistency and eliminate erroneous updates caused by imperfect supervision. Extensive experiments demonstrate superior results and in particular establishing new state-of-the-art performances with more consistent geometry and finer details.
comment: CVPR 2025
☆ Shaken, Not Stirred: A Novel Dataset for Visual Understanding of Glasses in Human-Robot Bartending Tasks IROS
Datasets for object detection often do not account for enough variety of glasses, due to their transparent and reflective properties. Specifically, open-vocabulary object detectors, widely used in embodied robotic agents, fail to distinguish subclasses of glasses. This scientific gap poses an issue to robotic applications that suffer from accumulating errors between detection, planning, and action execution. The paper introduces a novel method for the acquisition of real-world data from RGB-D sensors that minimizes human effort. We propose an auto-labeling pipeline that generates labels for all the acquired frames based on the depth measurements. We provide a novel real-world glass object dataset that was collected on the Neuro-Inspired COLlaborator (NICOL), a humanoid robot platform. The data set consists of 7850 images recorded from five different cameras. We show that our trained baseline model outperforms state-of-the-art open-vocabulary approaches. In addition, we deploy our baseline model in an embodied agent approach to the NICOL platform, on which it achieves a success rate of 81% in a human-robot bartending scenario.
comment: Submitted to IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2025
☆ ControlFill: Spatially Adjustable Image Inpainting from Prompt Learning
In this report, I present an inpainting framework named \textit{ControlFill}, which involves training two distinct prompts: one for generating plausible objects within a designated mask (\textit{creation}) and another for filling the region by extending the background (\textit{removal}). During the inference stage, these learned embeddings guide a diffusion network that operates without requiring heavy text encoders. By adjusting the relative significance of the two prompts and employing classifier-free guidance, users can control the intensity of removal or creation. Furthermore, I introduce a method to spatially vary the intensity of guidance by assigning different scales to individual pixels.
☆ TAIL: Text-Audio Incremental Learning
Many studies combine text and audio to capture multi-modal information but they overlook the model's generalization ability on new datasets. Introducing new datasets may affect the feature space of the original dataset, leading to catastrophic forgetting. Meanwhile, large model parameters can significantly impact training performance. To address these limitations, we introduce a novel task called Text-Audio Incremental Learning (TAIL) task for text-audio retrieval, and propose a new method, PTAT, Prompt Tuning for Audio-Text incremental learning. This method utilizes prompt tuning to optimize the model parameters while incorporating an audio-text similarity and feature distillation module to effectively mitigate catastrophic forgetting. We benchmark our method and previous incremental learning methods on AudioCaps, Clotho, BBC Sound Effects and Audioset datasets, and our method outperforms previous methods significantly, particularly demonstrating stronger resistance to forgetting on older datasets. Compared to the full-parameters Finetune (Sequential) method, our model only requires 2.42\% of its parameters, achieving 4.46\% higher performance.
comment: 4 figures, 5 tables
☆ How to Move Your Dragon: Text-to-Motion Synthesis for Large-Vocabulary Objects
Motion synthesis for diverse object categories holds great potential for 3D content creation but remains underexplored due to two key challenges: (1) the lack of comprehensive motion datasets that include a wide range of high-quality motions and annotations, and (2) the absence of methods capable of handling heterogeneous skeletal templates from diverse objects. To address these challenges, we contribute the following: First, we augment the Truebones Zoo dataset, a high-quality animal motion dataset covering over 70 species, by annotating it with detailed text descriptions, making it suitable for text-based motion synthesis. Second, we introduce rig augmentation techniques that generate diverse motion data while preserving consistent dynamics, enabling models to adapt to various skeletal configurations. Finally, we redesign existing motion diffusion models to dynamically adapt to arbitrary skeletal templates, enabling motion synthesis for a diverse range of objects with varying structures. Experiments show that our method learns to generate high-fidelity motions from textual descriptions for diverse and even unseen objects, setting a strong foundation for motion synthesis across diverse object categories and skeletal templates. Qualitative results are available on this link: t2m4lvo.github.io
☆ An Egocentric Vision-Language Model based Portable Real-time Smart Assistant
We present Vinci, a vision-language system designed to provide real-time, comprehensive AI assistance on portable devices. At its core, Vinci leverages EgoVideo-VL, a novel model that integrates an egocentric vision foundation model with a large language model (LLM), enabling advanced functionalities such as scene understanding, temporal grounding, video summarization, and future planning. To enhance its utility, Vinci incorporates a memory module for processing long video streams in real time while retaining contextual history, a generation module for producing visual action demonstrations, and a retrieval module that bridges egocentric and third-person perspectives to provide relevant how-to videos for skill acquisition. Unlike existing systems that often depend on specialized hardware, Vinci is hardware-agnostic, supporting deployment across a wide range of devices, including smartphones and wearable cameras. In our experiments, we first demonstrate the superior performance of EgoVideo-VL on multiple public benchmarks, showcasing its vision-language reasoning and contextual understanding capabilities. We then conduct a series of user studies to evaluate the real-world effectiveness of Vinci, highlighting its adaptability and usability in diverse scenarios. We hope Vinci can establish a new framework for portable, real-time egocentric AI systems, empowering users with contextual and actionable insights. Including the frontend, backend, and models, all codes of Vinci are available at https://github.com/OpenGVLab/vinci.
☆ Geometry-Constrained Monocular Scale Estimation Using Semantic Segmentation for Dynamic Scenes
Monocular visual localization plays a pivotal role in advanced driver assistance systems and autonomous driving by estimating a vehicle's ego-motion from a single pinhole camera. Nevertheless, conventional monocular visual odometry encoun-ters challenges in scale estimation due to the absence of depth information during projection. Previous methodologies, whether rooted in physical constraints or deep learning paradigms, con-tend with issues related to computational complexity and the management of dynamic objects. This study extends our prior research, presenting innovative strategies for ego-motion estima-tion and the selection of ground points. Striving for a nuanced equilibrium between computational efficiency and precision, we propose a hybrid method that leverages the SegNeXt model for real-time applications, encompassing both ego-motion estimation and ground point selection. Our methodology incorporates dy-namic object masks to eliminate unstable features and employs ground plane masks for meticulous triangulation. Furthermore, we exploit Geometry-constraint to delineate road regions for scale recovery. The integration of this approach with the mo-nocular version of ORB-SLAM3 culminates in the accurate esti-mation of a road model, a pivotal component in our scale recov-ery process. Rigorous experiments, conducted on the KITTI da-taset, systematically compare our method with existing monocu-lar visual odometry algorithms and contemporary scale recovery methodologies. The results undeniably confirm the superior ef-fectiveness of our approach, surpassing state-of-the-art visual odometry algorithms. Our source code is available at https://git hub.com/bFr0zNq/MVOSegScale.
☆ Synthetic Data is an Elegant GIFT for Continual Vision-Language Models CVPR 2025
Pre-trained Vision-Language Models (VLMs) require Continual Learning (CL) to efficiently update their knowledge and adapt to various downstream tasks without retraining from scratch. However, for VLMs, in addition to the loss of knowledge previously learned from downstream tasks, pre-training knowledge is also corrupted during continual fine-tuning. This issue is exacerbated by the unavailability of original pre-training data, leaving VLM's generalization ability degrading. In this paper, we propose GIFT, a novel continual fine-tuning approach that utilizes synthetic data to overcome catastrophic forgetting in VLMs. Taking advantage of recent advances in text-to-image synthesis, we employ a pre-trained diffusion model to recreate both pre-training and learned downstream task data. In this way, the VLM can revisit previous knowledge through distillation on matching diffusion-generated images and corresponding text prompts. Leveraging the broad distribution and high alignment between synthetic image-text pairs in VLM's feature space, we propose a contrastive distillation loss along with an image-text alignment constraint. To further combat in-distribution overfitting and enhance distillation performance with limited amount of generated data, we incorporate adaptive weight consolidation, utilizing Fisher information from these synthetic image-text pairs and achieving a better stability-plasticity balance. Extensive experiments demonstrate that our method consistently outperforms previous state-of-the-art approaches across various settings.
comment: This work is accepted by CVPR 2025. Modifications may be performed
☆ Spiking Meets Attention: Efficient Remote Sensing Image Super-Resolution with Attention Spiking Neural Networks
Spiking neural networks (SNNs) are emerging as a promising alternative to traditional artificial neural networks (ANNs), offering biological plausibility and energy efficiency. Despite these merits, SNNs are frequently hampered by limited capacity and insufficient representation power, yet remain underexplored in remote sensing super-resolution (SR) tasks. In this paper, we first observe that spiking signals exhibit drastic intensity variations across diverse textures, highlighting an active learning state of the neurons. This observation motivates us to apply SNNs for efficient SR of RSIs. Inspired by the success of attention mechanisms in representing salient information, we devise the spiking attention block (SAB), a concise yet effective component that optimizes membrane potentials through inferred attention weights, which, in turn, regulates spiking activity for superior feature representation. Our key contributions include: 1) we bridge the independent modulation between temporal and channel dimensions, facilitating joint feature correlation learning, and 2) we access the global self-similar patterns in large-scale remote sensing imagery to infer spatial attention weights, incorporating effective priors for realistic and faithful reconstruction. Building upon SAB, we proposed SpikeSR, which achieves state-of-the-art performance across various remote sensing benchmarks such as AID, DOTA, and DIOR, while maintaining high computational efficiency. The code of SpikeSR will be available upon paper acceptance.
☆ Energy-Guided Optimization for Personalized Image Editing with Pretrained Text-to-Image Diffusion Models
The rapid advancement of pretrained text-driven diffusion models has significantly enriched applications in image generation and editing. However, as the demand for personalized content editing increases, new challenges emerge especially when dealing with arbitrary objects and complex scenes. Existing methods usually mistakes mask as the object shape prior, which struggle to achieve a seamless integration result. The mostly used inversion noise initialization also hinders the identity consistency towards the target object. To address these challenges, we propose a novel training-free framework that formulates personalized content editing as the optimization of edited images in the latent space, using diffusion models as the energy function guidance conditioned by reference text-image pairs. A coarse-to-fine strategy is proposed that employs text energy guidance at the early stage to achieve a natural transition toward the target class and uses point-to-point feature-level image energy guidance to perform fine-grained appearance alignment with the target object. Additionally, we introduce the latent space content composition to enhance overall identity consistency with the target. Extensive experiments demonstrate that our method excels in object replacement even with a large domain gap, highlighting its potential for high-quality, personalized image editing.
☆ Bridging the Vision-Brain Gap with an Uncertainty-Aware Blur Prior
Can our brain signals faithfully reflect the original visual stimuli, even including high-frequency details? Although human perceptual and cognitive capacities enable us to process and remember visual information, these abilities are constrained by several factors, such as limited attentional resources and the finite capacity of visual memory. When visual stimuli are processed by human visual system into brain signals, some information is inevitably lost, leading to a discrepancy known as the \textbf{System GAP}. Additionally, perceptual and cognitive dynamics, along with technical noise in signal acquisition, degrade the fidelity of brain signals relative to the visual stimuli, known as the \textbf{Random GAP}. When encoded brain representations are directly aligned with the corresponding pretrained image features, the System GAP and Random GAP between paired data challenge the model, requiring it to bridge these gaps. However, in the context of limited paired data, these gaps are difficult for the model to learn, leading to overfitting and poor generalization to new data. To address these GAPs, we propose a simple yet effective approach called the \textbf{Uncertainty-aware Blur Prior (UBP)}. It estimates the uncertainty within the paired data, reflecting the mismatch between brain signals and visual stimuli. Based on this uncertainty, UBP dynamically blurs the high-frequency details of the original images, reducing the impact of the mismatch and improving alignment. Our method achieves a top-1 accuracy of \textbf{50.9\%} and a top-5 accuracy of \textbf{79.7\%} on the zero-shot brain-to-image retrieval task, surpassing previous state-of-the-art methods by margins of \textbf{13.7\%} and \textbf{9.8\%}, respectively. Code is available at \href{https://github.com/HaitaoWuTJU/Uncertainty-aware-Blur-Prior}{GitHub}.
☆ Learning 3D Medical Image Models From Brain Functional Connectivity Network Supervision For Mental Disorder Diagnosis
In MRI-based mental disorder diagnosis, most previous studies focus on functional connectivity network (FCN) derived from functional MRI (fMRI). However, the small size of annotated fMRI datasets restricts its wide application. Meanwhile, structural MRIs (sMRIs), such as 3D T1-weighted (T1w) MRI, which are commonly used and readily accessible in clinical settings, are often overlooked. To integrate the complementary information from both function and structure for improved diagnostic accuracy, we propose CINP (Contrastive Image-Network Pre-training), a framework that employs contrastive learning between sMRI and FCN. During pre-training, we incorporate masked image modeling and network-image matching to enhance visual representation learning and modality alignment. Since the CINP facilitates knowledge transfer from FCN to sMRI, we introduce network prompting. It utilizes only sMRI from suspected patients and a small amount of FCNs from different patient classes for diagnosing mental disorders, which is practical in real-world clinical scenario. The competitive performance on three mental disorder diagnosis tasks demonstrate the effectiveness of the CINP in integrating multimodal MRI information, as well as the potential of incorporating sMRI into clinical diagnosis using network prompting.
☆ FUSE: First-Order and Second-Order Unified SynthEsis in Stochastic Optimization
Stochastic optimization methods have actively been playing a critical role in modern machine learning algorithms to deliver decent performance. While numerous works have proposed and developed diverse approaches, first-order and second-order methods are in entirely different situations. The former is significantly pivotal and dominating in emerging deep learning but only leads convergence to a stationary point. However, second-order methods are less popular due to their computational intensity in large-dimensional problems. This paper presents a novel method that leverages both the first-order and second-order methods in a unified algorithmic framework, termed FUSE, from which a practical version (PV) is derived accordingly. FUSE-PV stands as a simple yet efficient optimization method involving a switch-over between first and second orders. Additionally, we develop different criteria that determine when to switch. FUSE-PV has provably shown a smaller computational complexity than SGD and Adam. To validate our proposed scheme, we present an ablation study on several simple test functions and show a comparison with baselines for benchmark datasets.
comment: 6 pages, 7 figures
☆ MASTER: Multimodal Segmentation with Text Prompts
RGB-Thermal fusion is a potential solution for various weather and light conditions in challenging scenarios. However, plenty of studies focus on designing complex modules to fuse different modalities. With the widespread application of large language models (LLMs), valuable information can be more effectively extracted from natural language. Therefore, we aim to leverage the advantages of large language models to design a structurally simple and highly adaptable multimodal fusion model architecture. We proposed MultimodAl Segmentation with TExt PRompts (MASTER) architecture, which integrates LLM into the fusion of RGB-Thermal multimodal data and allows complex query text to participate in the fusion process. Our model utilizes a dual-path structure to extract information from different modalities of images. Additionally, we employ LLM as the core module for multimodal fusion, enabling the model to generate learnable codebook tokens from RGB, thermal images, and textual information. A lightweight image decoder is used to obtain semantic segmentation results. The proposed MASTER performs exceptionally well in benchmark tests across various automated driving scenarios, yielding promising results.
☆ Conformal forecasting for surgical instrument trajectory
Forecasting surgical instrument trajectories and predicting the next surgical action recently started to attract attention from the research community. Both these tasks are crucial for automation and assistance in endoscopy surgery. Given the safety-critical nature of these tasks, reliable uncertainty quantification is essential. Conformal prediction is a fast-growing and widely recognized framework for uncertainty estimation in machine learning and computer vision, offering distribution-free, theoretically valid prediction intervals. In this work, we explore the application of standard conformal prediction and conformalized quantile regression to estimate uncertainty in forecasting surgical instrument motion, i.e., predicting direction and magnitude of surgical instruments' future motion. We analyze and compare their coverage and interval sizes, assessing the impact of multiple hypothesis testing and correction methods. Additionally, we show how these techniques can be employed to produce useful uncertainty heatmaps. To the best of our knowledge, this is the first study applying conformal prediction to surgical guidance, marking an initial step toward constructing principled prediction intervals with formal coverage guarantees in this domain.
☆ DuCos: Duality Constrained Depth Super-Resolution via Foundation Model
We introduce DuCos, a novel depth super-resolution framework grounded in Lagrangian duality theory, offering a flexible integration of multiple constraints and reconstruction objectives to enhance accuracy and robustness. Our DuCos is the first to significantly improve generalization across diverse scenarios with foundation models as prompts. The prompt design consists of two key components: Correlative Fusion (CF) and Gradient Regulation (GR). CF facilitates precise geometric alignment and effective fusion between prompt and depth features, while GR refines depth predictions by enforcing consistency with sharp-edged depth maps derived from foundation models. Crucially, these prompts are seamlessly embedded into the Lagrangian constraint term, forming a synergistic and principled framework. Extensive experiments demonstrate that DuCos outperforms existing state-of-the-art methods, achieving superior accuracy, robustness, and generalization. The source codes and pre-trained models will be publicly available.
☆ The Role of Visual Modality in Multimodal Mathematical Reasoning: Challenges and Insights
Recent research has increasingly focused on multimodal mathematical reasoning, particularly emphasizing the creation of relevant datasets and benchmarks. Despite this, the role of visual information in reasoning has been underexplored. Our findings show that existing multimodal mathematical models minimally leverage visual information, and model performance remains largely unaffected by changes to or removal of images in the dataset. We attribute this to the dominance of textual information and answer options that inadvertently guide the model to correct answers. To improve evaluation methods, we introduce the HC-M3D dataset, specifically designed to require image reliance for problem-solving and to challenge models with similar, yet distinct, images that change the correct answer. In testing leading models, their failure to detect these subtle visual differences suggests limitations in current visual perception capabilities. Additionally, we observe that the common approach of improving general VQA capabilities by combining various types of image encoders does not contribute to math reasoning performance. This finding also presents a challenge to enhancing visual reliance during math reasoning. Our benchmark and code would be available at \href{https://github.com/Yufang-Liu/visual_modality_role}{https://github.com/Yufang-Liu/visual\_modality\_role}.
☆ WeakSupCon: Weakly Supervised Contrastive Learning for Encoder Pre-training
Weakly supervised multiple instance learning (MIL) is a challenging task given that only bag-level labels are provided, while each bag typically contains multiple instances. This topic has been extensively studied in histopathological image analysis, where labels are usually available only at the whole slide image (WSI) level, while each whole slide image can be divided into thousands of small image patches for training. The dominant MIL approaches take fixed patch features as inputs to address computational constraints and ensure model stability. These features are commonly generated by encoders pre-trained on ImageNet, foundation encoders pre-trained on large datasets, or through self-supervised learning on local datasets. While the self-supervised encoder pre-training on the same dataset as downstream MIL tasks helps mitigate domain shift and generate better features, the bag-level labels are not utilized during the process, and the features of patches from different categories may cluster together, reducing classification performance on MIL tasks. Recently, pre-training with supervised contrastive learning (SupCon) has demonstrated superior performance compared to self-supervised contrastive learning and even end-to-end training on traditional image classification tasks. In this paper, we propose a novel encoder pre-training method for downstream MIL tasks called Weakly Supervised Contrastive Learning (WeakSupCon) that utilizes bag-level labels. In our method, we employ multi-task learning and define distinct contrastive learning losses for samples with different bag labels. Our experiments demonstrate that the features generated using WeakSupCon significantly enhance MIL classification performance compared to self-supervised approaches across three datasets.
☆ CA-W3D: Leveraging Context-Aware Knowledge for Weakly Supervised Monocular 3D Detection
Weakly supervised monocular 3D detection, while less annotation-intensive, often struggles to capture the global context required for reliable 3D reasoning. Conventional label-efficient methods focus on object-centric features, neglecting contextual semantic relationships that are critical in complex scenes. In this work, we propose a Context-Aware Weak Supervision for Monocular 3D object detection, namely CA-W3D, to address this limitation in a two-stage training paradigm. Specifically, we first introduce a pre-training stage employing Region-wise Object Contrastive Matching (ROCM), which aligns regional object embeddings derived from a trainable monocular 3D encoder and a frozen open-vocabulary 2D visual grounding model. This alignment encourages the monocular encoder to discriminate scene-specific attributes and acquire richer contextual knowledge. In the second stage, we incorporate a pseudo-label training process with a Dual-to-One Distillation (D2OD) mechanism, which effectively transfers contextual priors into the monocular encoder while preserving spatial fidelity and maintaining computational efficiency during inference. Extensive experiments conducted on the public KITTI benchmark demonstrate the effectiveness of our approach, surpassing the SoTA method over all metrics, highlighting the importance of contextual-aware knowledge in weakly-supervised monocular 3D detection.
comment: The paper includes 8 pages, 6 figures and 4 tables
☆ Robust Multi-View Learning via Representation Fusion of Sample-Level Attention and Alignment of Simulated Perturbation
Recently, multi-view learning (MVL) has garnered significant attention due to its ability to fuse discriminative information from multiple views. However, real-world multi-view datasets are often heterogeneous and imperfect, which usually makes MVL methods designed for specific combinations of views lack application potential and limits their effectiveness. To address this issue, we propose a novel robust MVL method (namely RML) with simultaneous representation fusion and alignment. Specifically, we introduce a simple yet effective multi-view transformer fusion network where we transform heterogeneous multi-view data into homogeneous word embeddings, and then integrate multiple views by the sample-level attention mechanism to obtain a fused representation. Furthermore, we propose a simulated perturbation based multi-view contrastive learning framework that dynamically generates the noise and unusable perturbations for simulating imperfect data conditions. The simulated noisy and unusable data obtain two distinct fused representations, and we utilize contrastive learning to align them for learning discriminative and robust representations. Our RML is self-supervised and can also be applied for downstream tasks as a regularization. In experiments, we employ it in unsupervised multi-view clustering, noise-label classification, and as a plug-and-play module for cross-modal hashing retrieval. Extensive comparison experiments and ablation studies validate the effectiveness of RML.
☆ DM-Adapter: Domain-Aware Mixture-of-Adapters for Text-Based Person Retrieval AAAI 2025
Text-based person retrieval (TPR) has gained significant attention as a fine-grained and challenging task that closely aligns with practical applications. Tailoring CLIP to person domain is now a emerging research topic due to the abundant knowledge of vision-language pretraining, but challenges still remain during fine-tuning: (i) Previous full-model fine-tuning in TPR is computationally expensive and prone to overfitting.(ii) Existing parameter-efficient transfer learning (PETL) for TPR lacks of fine-grained feature extraction. To address these issues, we propose Domain-Aware Mixture-of-Adapters (DM-Adapter), which unifies Mixture-of-Experts (MOE) and PETL to enhance fine-grained feature representations while maintaining efficiency. Specifically, Sparse Mixture-of-Adapters is designed in parallel to MLP layers in both vision and language branches, where different experts specialize in distinct aspects of person knowledge to handle features more finely. To promote the router to exploit domain information effectively and alleviate the routing imbalance, Domain-Aware Router is then developed by building a novel gating function and injecting learnable domain-aware prompts. Extensive experiments show that our DM-Adapter achieves state-of-the-art performance, outperforming previous methods by a significant margin.
comment: 9 pages, 5 figures, accepted by AAAI 2025
☆ Robust Computer-Vision based Construction Site Detection for Assistive-Technology Applications
Navigating urban environments poses significant challenges for people with disabilities, particularly those with blindness and low vision. Environments with dynamic and unpredictable elements like construction sites are especially challenging. Construction sites introduce hazards like uneven surfaces, obstructive barriers, hazardous materials, and excessive noise, and they can alter routing, complicating safe mobility. Existing assistive technologies are limited, as navigation apps do not account for construction sites during trip planning, and detection tools that attempt hazard recognition struggle to address the extreme variability of construction paraphernalia. This study introduces a novel computer vision-based system that integrates open-vocabulary object detection, a YOLO-based scaffolding-pole detection model, and an optical character recognition (OCR) module to comprehensively identify and interpret construction site elements for assistive navigation. In static testing across seven construction sites, the system achieved an overall accuracy of 88.56\%, reliably detecting objects from 2m to 10m within a 0$^\circ$ -- 75$^\circ$ angular offset. At closer distances (2--4m), the detection rate was 100\% at all tested angles. At
☆ Real-time Spatial-temporal Traversability Assessment via Feature-based Sparse Gaussian Process
Terrain analysis is critical for the practical application of ground mobile robots in real-world tasks, especially in outdoor unstructured environments. In this paper, we propose a novel spatial-temporal traversability assessment method, which aims to enable autonomous robots to effectively navigate through complex terrains. Our approach utilizes sparse Gaussian processes (SGP) to extract geometric features (curvature, gradient, elevation, etc.) directly from point cloud scans. These features are then used to construct a high-resolution local traversability map. Then, we design a spatial-temporal Bayesian Gaussian kernel (BGK) inference method to dynamically evaluate traversability scores, integrating historical and real-time data while considering factors such as slope, flatness, gradient, and uncertainty metrics. GPU acceleration is applied in the feature extraction step, and the system achieves real-time performance. Extensive simulation experiments across diverse terrain scenarios demonstrate that our method outperforms SOTA approaches in both accuracy and computational efficiency. Additionally, we develop an autonomous navigation framework integrated with the traversability map and validate it with a differential driven vehicle in complex outdoor environments. Our code will be open-source for further research and development by the community, https://github.com/ZJU-FAST-Lab/FSGP_BGK.
comment: 8 pages, 10 figures
☆ Q-PART: Quasi-Periodic Adaptive Regression with Test-time Training for Pediatric Left Ventricular Ejection Fraction Regression CVPR 2025
In this work, we address the challenge of adaptive pediatric Left Ventricular Ejection Fraction (LVEF) assessment. While Test-time Training (TTT) approaches show promise for this task, they suffer from two significant limitations. Existing TTT works are primarily designed for classification tasks rather than continuous value regression, and they lack mechanisms to handle the quasi-periodic nature of cardiac signals. To tackle these issues, we propose a novel \textbf{Q}uasi-\textbf{P}eriodic \textbf{A}daptive \textbf{R}egression with \textbf{T}est-time Training (Q-PART) framework. In the training stage, the proposed Quasi-Period Network decomposes the echocardiogram into periodic and aperiodic components within latent space by combining parameterized helix trajectories with Neural Controlled Differential Equations. During inference, our framework further employs a variance minimization strategy across image augmentations that simulate common quality issues in echocardiogram acquisition, along with differential adaptation rates for periodic and aperiodic components. Theoretical analysis is provided to demonstrate that our variance minimization objective effectively bounds the regression error under mild conditions. Furthermore, extensive experiments across three pediatric age groups demonstrate that Q-PART not only significantly outperforms existing approaches in pediatric LVEF prediction, but also exhibits strong clinical screening capability with high mAUROC scores (up to 0.9747) and maintains gender-fair performance across all metrics, validating its robustness and practical utility in pediatric echocardiography analysis.
comment: Accepted to CVPR 2025
☆ Token-Efficient Long Video Understanding for Multimodal LLMs
Recent advances in video-based multimodal large language models (Video-LLMs) have significantly improved video understanding by processing videos as sequences of image frames. However, many existing methods treat frames independently in the vision backbone, lacking explicit temporal modeling, which limits their ability to capture dynamic patterns and efficiently handle long videos. To address these limitations, we introduce STORM (\textbf{S}patiotemporal \textbf{TO}ken \textbf{R}eduction for \textbf{M}ultimodal LLMs), a novel architecture incorporating a dedicated temporal encoder between the image encoder and the LLM. Our temporal encoder leverages the Mamba State Space Model to integrate temporal information into image tokens, generating enriched representations that preserve inter-frame dynamics across the entire video sequence. This enriched encoding not only enhances video reasoning capabilities but also enables effective token reduction strategies, including test-time sampling and training-based temporal and spatial pooling, substantially reducing computational demands on the LLM without sacrificing key temporal information. By integrating these techniques, our approach simultaneously reduces training and inference latency while improving performance, enabling efficient and robust video understanding over extended temporal contexts. Extensive evaluations show that STORM achieves state-of-the-art results across various long video understanding benchmarks (more than 5\% improvement on MLVU and LongVideoBench) while reducing the computation costs by up to $8\times$ and the decoding latency by 2.4-2.9$\times$ for the fixed numbers of input frames. Project page is available at https://research.nvidia.com/labs/lpr/storm
☆ Diff-Reg v2: Diffusion-Based Matching Matrix Estimation for Image Matching and 3D Registration
Establishing reliable correspondences is crucial for all registration tasks, including 2D image registration, 3D point cloud registration, and 2D-3D image-to-point cloud registration. However, these tasks are often complicated by challenges such as scale inconsistencies, symmetry, and large deformations, which can lead to ambiguous matches. Previous feature-based and correspondence-based methods typically rely on geometric or semantic features to generate or polish initial potential correspondences. Some methods typically leverage specific geometric priors, such as topological preservation, to devise diverse and innovative strategies tailored to a given enhancement goal, which cannot be exhaustively enumerated. Additionally, many previous approaches rely on a single-step prediction head, which can struggle with local minima in complex matching scenarios. To address these challenges, we introduce an innovative paradigm that leverages a diffusion model in matrix space for robust matching matrix estimation. Our model treats correspondence estimation as a denoising diffusion process in the matching matrix space, gradually refining the intermediate matching matrix to the optimal one. Specifically, we apply the diffusion model in the doubly stochastic matrix space for 3D-3D and 2D-3D registration tasks. In the 2D image registration task, we deploy the diffusion model in a matrix subspace where dual-softmax projection regularization is applied. For all three registration tasks, we provide adaptive matching matrix embedding implementations tailored to the specific characteristics of each task while maintaining a consistent "match-to-warp" encoding pattern. Furthermore, we adopt a lightweight design for the denoising module. In inference, once points or image features are extracted and fixed, this module performs multi-step denoising predictions through reverse sampling.
comment: arXiv admin note: text overlap with arXiv:2403.19919
☆ DVM-SLAM: Decentralized Visual Monocular Simultaneous Localization and Mapping for Multi-Agent Systems
Cooperative Simultaneous Localization and Mapping (C-SLAM) enables multiple agents to work together in mapping unknown environments while simultaneously estimating their own positions. This approach enhances robustness, scalability, and accuracy by sharing information between agents, reducing drift, and enabling collective exploration of larger areas. In this paper, we present Decentralized Visual Monocular SLAM (DVM-SLAM), the first open-source decentralized monocular C-SLAM system. By only utilizing low-cost and light-weight monocular vision sensors, our system is well suited for small robots and micro aerial vehicles (MAVs). DVM-SLAM's real-world applicability is validated on physical robots with a custom collision avoidance framework, showcasing its potential in real-time multi-agent autonomous navigation scenarios. We also demonstrate comparable accuracy to state-of-the-art centralized monocular C-SLAM systems. We open-source our code and provide supplementary material online.
☆ GAGrasp: Geometric Algebra Diffusion for Dexterous Grasping ICRA 2025
We propose GAGrasp, a novel framework for dexterous grasp generation that leverages geometric algebra representations to enforce equivariance to SE(3) transformations. By encoding the SE(3) symmetry constraint directly into the architecture, our method improves data and parameter efficiency while enabling robust grasp generation across diverse object poses. Additionally, we incorporate a differentiable physics-informed refinement layer, which ensures that generated grasps are physically plausible and stable. Extensive experiments demonstrate the model's superior performance in generalization, stability, and adaptability compared to existing methods. Additional details at https://gagrasp.github.io/
comment: Accepted at ICRA 2025
☆ Simple Self Organizing Map with Visual Transformer
Vision Transformers (ViTs) have demonstrated exceptional performance in various vision tasks. However, they tend to underperform on smaller datasets due to their inherent lack of inductive biases. Current approaches address this limitation implicitly-often by pairing ViTs with pretext tasks or by distilling knowledge from convolutional neural networks (CNNs) to strengthen the prior. In contrast, Self-Organizing Maps (SOMs), a widely adopted self-supervised framework, are inherently structured to preserve topology and spatial organization, making them a promising candidate to directly address the limitations of ViTs in limited or small training datasets. Despite this potential, equipping SOMs with modern deep learning architectures remains largely unexplored. In this study, we conduct a novel exploration on how Vision Transformers (ViTs) and Self-Organizing Maps (SOMs) can empower each other, aiming to bridge this critical research gap. Our findings demonstrate that these architectures can synergistically enhance each other, leading to significantly improved performance in both unsupervised and supervised tasks. Code will be publicly available.
comment: 5 pages, 4 figures. Submitted to IEEE. All experiments and code work were performed by the first author, with the second author serving in a PI/mentor role, guiding the progression of the work
☆ SCSA: A Plug-and-Play Semantic Continuous-Sparse Attention for Arbitrary Semantic Style Transfer CVPR 2025
Attention-based arbitrary style transfer methods, including CNN-based, Transformer-based, and Diffusion-based, have flourished and produced high-quality stylized images. However, they perform poorly on the content and style images with the same semantics, i.e., the style of the corresponding semantic region of the generated stylized image is inconsistent with that of the style image. We argue that the root cause lies in their failure to consider the relationship between local regions and semantic regions. To address this issue, we propose a plug-and-play semantic continuous-sparse attention, dubbed SCSA, for arbitrary semantic style transfer -- each query point considers certain key points in the corresponding semantic region. Specifically, semantic continuous attention ensures each query point fully attends to all the continuous key points in the same semantic region that reflect the overall style characteristics of that region; Semantic sparse attention allows each query point to focus on the most similar sparse key point in the same semantic region that exhibits the specific stylistic texture of that region. By combining the two modules, the resulting SCSA aligns the overall style of the corresponding semantic regions while transferring the vivid textures of these regions. Qualitative and quantitative results prove that SCSA enables attention-based arbitrary style transfer methods to produce high-quality semantic stylized images.
comment: Accepted by CVPR 2025
☆ Fractional Correspondence Framework in Detection Transformer
The Detection Transformer (DETR), by incorporating the Hungarian algorithm, has significantly simplified the matching process in object detection tasks. This algorithm facilitates optimal one-to-one matching of predicted bounding boxes to ground-truth annotations during training. While effective, this strict matching process does not inherently account for the varying densities and distributions of objects, leading to suboptimal correspondences such as failing to handle multiple detections of the same object or missing small objects. To address this, we propose the Regularized Transport Plan (RTP). RTP introduces a flexible matching strategy that captures the cost of aligning predictions with ground truths to find the most accurate correspondences between these sets. By utilizing the differentiable Sinkhorn algorithm, RTP allows for soft, fractional matching rather than strict one-to-one assignments. This approach enhances the model's capability to manage varying object densities and distributions effectively. Our extensive evaluations on the MS-COCO and VOC benchmarks demonstrate the effectiveness of our approach. RTP-DETR, surpassing the performance of the Deform-DETR and the recently introduced DINO-DETR, achieving absolute gains in mAP of +3.8% and +1.7%, respectively.
☆ WeakMedSAM: Weakly-Supervised Medical Image Segmentation via SAM with Sub-Class Exploration and Prompt Affinity Mining
We have witnessed remarkable progress in foundation models in vision tasks. Currently, several recent works have utilized the segmenting anything model (SAM) to boost the segmentation performance in medical images, where most of them focus on training an adaptor for fine-tuning a large amount of pixel-wise annotated medical images following a fully supervised manner. In this paper, to reduce the labeling cost, we investigate a novel weakly-supervised SAM-based segmentation model, namely WeakMedSAM. Specifically, our proposed WeakMedSAM contains two modules: 1) to mitigate severe co-occurrence in medical images, a sub-class exploration module is introduced to learn accurate feature representations. 2) to improve the quality of the class activation maps, our prompt affinity mining module utilizes the prompt capability of SAM to obtain an affinity map for random-walk refinement. Our method can be applied to any SAM-like backbone, and we conduct experiments with SAMUS and EfficientSAM. The experimental results on three popularly-used benchmark datasets, i.e., BraTS 2019, AbdomenCT-1K, and MSD Cardiac dataset, show the promising results of our proposed WeakMedSAM. Our code is available at https://github.com/wanghr64/WeakMedSAM.
☆ Image-Based Relocalization and Alignment for Long-Term Monitoring of Dynamic Underwater Environments
Effective monitoring of underwater ecosystems is crucial for tracking environmental changes, guiding conservation efforts, and ensuring long-term ecosystem health. However, automating underwater ecosystem management with robotic platforms remains challenging due to the complexities of underwater imagery, which pose significant difficulties for traditional visual localization methods. We propose an integrated pipeline that combines Visual Place Recognition (VPR), feature matching, and image segmentation on video-derived images. This method enables robust identification of revisited areas, estimation of rigid transformations, and downstream analysis of ecosystem changes. Furthermore, we introduce the SQUIDLE+ VPR Benchmark-the first large-scale underwater VPR benchmark designed to leverage an extensive collection of unstructured data from multiple robotic platforms, spanning time intervals from days to years. The dataset encompasses diverse trajectories, arbitrary overlap and diverse seafloor types captured under varying environmental conditions, including differences in depth, lighting, and turbidity. Our code is available at: https://github.com/bev-gorry/underloc
☆ Brain Tumor Detection in MRI Based on Federated Learning with YOLOv11
One of the primary challenges in medical diagnostics is the accurate and efficient use of magnetic resonance imaging (MRI) for the detection of brain tumors. But the current machine learning (ML) approaches have two major limitations, data privacy and high latency. To solve the problem, in this work we propose a federated learning architecture for a better accurate brain tumor detection incorporating the YOLOv11 algorithm. In contrast to earlier methods of centralized learning, our federated learning approach protects the underlying medical data while supporting cooperative deep learning model training across multiple institutions. To allow the YOLOv11 model to locate and identify tumor areas, we adjust it to handle MRI data. To ensure robustness and generalizability, the model is trained and tested on a wide range of MRI data collected from several anonymous medical facilities. The results indicate that our method significantly maintains higher accuracy than conventional approaches.
☆ Instrument-Splatting: Controllable Photorealistic Reconstruction of Surgical Instruments Using Gaussian Splatting
Real2Sim is becoming increasingly important with the rapid development of surgical artificial intelligence (AI) and autonomy. In this work, we propose a novel Real2Sim methodology, \textit{Instrument-Splatting}, that leverages 3D Gaussian Splatting to provide fully controllable 3D reconstruction of surgical instruments from monocular surgical videos. To maintain both high visual fidelity and manipulability, we introduce a geometry pre-training to bind Gaussian point clouds on part mesh with accurate geometric priors and define a forward kinematics to control the Gaussians as flexible as real instruments. Afterward, to handle unposed videos, we design a novel instrument pose tracking method leveraging semantics-embedded Gaussians to robustly refine per-frame instrument poses and joint states in a render-and-compare manner, which allows our instrument Gaussian to accurately learn textures and reach photorealistic rendering. We validated our method on 2 publicly released surgical videos and 4 videos collected on ex vivo tissues and green screens. Quantitative and qualitative evaluations demonstrate the effectiveness and superiority of the proposed method.
comment: 11 pages, 5 figures
☆ Surgical Gaussian Surfels: Highly Accurate Real-time Surgical Scene Rendering
Accurate geometric reconstruction of deformable tissues in monocular endoscopic video remains a fundamental challenge in robot-assisted minimally invasive surgery. Although recent volumetric and point primitive methods based on neural radiance fields (NeRF) and 3D Gaussian primitives have efficiently rendered surgical scenes, they still struggle with handling artifact-free tool occlusions and preserving fine anatomical details. These limitations stem from unrestricted Gaussian scaling and insufficient surface alignment constraints during reconstruction. To address these issues, we introduce Surgical Gaussian Surfels (SGS), which transforms anisotropic point primitives into surface-aligned elliptical splats by constraining the scale component of the Gaussian covariance matrix along the view-aligned axis. We predict accurate surfel motion fields using a lightweight Multi-Layer Perceptron (MLP) coupled with locality constraints to handle complex tissue deformations. We use homodirectional view-space positional gradients to capture fine image details by splitting Gaussian Surfels in over-reconstructed regions. In addition, we define surface normals as the direction of the steepest density change within each Gaussian surfel primitive, enabling accurate normal estimation without requiring monocular normal priors. We evaluate our method on two in-vivo surgical datasets, where it outperforms current state-of-the-art methods in surface geometry, normal map quality, and rendering efficiency, while remaining competitive in real-time rendering performance. We make our code available at https://github.com/aloma85/SurgicalGaussianSurfels
☆ Spatial-Temporal Perception with Causal Inference for Naturalistic Driving Action Recognition
Naturalistic driving action recognition is essential for vehicle cabin monitoring systems. However, the complexity of real-world backgrounds presents significant challenges for this task, and previous approaches have struggled with practical implementation due to their limited ability to observe subtle behavioral differences and effectively learn inter-frame features from video. In this paper, we propose a novel Spatial-Temporal Perception (STP) architecture that emphasizes both temporal information and spatial relationships between key objects, incorporating a causal decoder to perform behavior recognition and temporal action localization. Without requiring multimodal input, STP directly extracts temporal and spatial distance features from RGB video clips. Subsequently, these dual features are jointly encoded by maximizing the expected likelihood across all possible permutations of the factorization order. By integrating temporal and spatial features at different scales, STP can perceive subtle behavioral changes in challenging scenarios. Additionally, we introduce a causal-aware module to explore relationships between video frame features, significantly enhancing detection efficiency and performance. We validate the effectiveness of our approach using two publicly available driver distraction detection benchmarks. The results demonstrate that our framework achieves state-of-the-art performance.
☆ FREAK: Frequency-modulated High-fidelity and Real-time Audio-driven Talking Portrait Synthesis
Achieving high-fidelity lip-speech synchronization in audio-driven talking portrait synthesis remains challenging. While multi-stage pipelines or diffusion models yield high-quality results, they suffer from high computational costs. Some approaches perform well on specific individuals with low resources, yet still exhibit mismatched lip movements. The aforementioned methods are modeled in the pixel domain. We observed that there are noticeable discrepancies in the frequency domain between the synthesized talking videos and natural videos. Currently, no research on talking portrait synthesis has considered this aspect. To address this, we propose a FREquency-modulated, high-fidelity, and real-time Audio-driven talKing portrait synthesis framework, named FREAK, which models talking portraits from the frequency domain perspective, enhancing the fidelity and naturalness of the synthesized portraits. FREAK introduces two novel frequency-based modules: 1) the Visual Encoding Frequency Modulator (VEFM) to couple multi-scale visual features in the frequency domain, better preserving visual frequency information and reducing the gap in the frequency spectrum between synthesized and natural frames. and 2) the Audio Visual Frequency Modulator (AVFM) to help the model learn the talking pattern in the frequency domain and improve audio-visual synchronization. Additionally, we optimize the model in both pixel domain and frequency domain jointly. Furthermore, FREAK supports seamless switching between one-shot and video dubbing settings, offering enhanced flexibility. Due to its superior performance, it can simultaneously support high-resolution video results and real-time inference. Extensive experiments demonstrate that our method synthesizes high-fidelity talking portraits with detailed facial textures and precise lip synchronization in real-time, outperforming state-of-the-art methods.
☆ PP-DocBee: Improving Multimodal Document Understanding Through a Bag of Tricks
With the rapid advancement of digitalization, various document images are being applied more extensively in production and daily life, and there is an increasingly urgent need for fast and accurate parsing of the content in document images. Therefore, this report presents PP-DocBee, a novel multimodal large language model designed for end-to-end document image understanding. First, we develop a data synthesis strategy tailored to document scenarios in which we build a diverse dataset to improve the model generalization. Then, we apply a few training techniques, including dynamic proportional sampling, data preprocessing, and OCR postprocessing strategies. Extensive evaluations demonstrate the superior performance of PP-DocBee, achieving state-of-the-art results on English document understanding benchmarks and even outperforming existing open source and commercial models in Chinese document understanding. The source code and pre-trained models are publicly available at \href{https://github.com/PaddlePaddle/PaddleMIX}{https://github.com/PaddlePaddle/PaddleMIX}.
☆ H3O: Hyper-Efficient 3D Occupancy Prediction with Heterogeneous Supervision ICRA 2025
3D occupancy prediction has recently emerged as a new paradigm for holistic 3D scene understanding and provides valuable information for downstream planning in autonomous driving. Most existing methods, however, are computationally expensive, requiring costly attention-based 2D-3D transformation and 3D feature processing. In this paper, we present a novel 3D occupancy prediction approach, H3O, which features highly efficient architecture designs that incur a significantly lower computational cost as compared to the current state-of-the-art methods. In addition, to compensate for the ambiguity in ground-truth 3D occupancy labels, we advocate leveraging auxiliary tasks to complement the direct 3D supervision. In particular, we integrate multi-camera depth estimation, semantic segmentation, and surface normal estimation via differentiable volume rendering, supervised by corresponding 2D labels that introduces rich and heterogeneous supervision signals. We conduct extensive experiments on the Occ3D-nuScenes and SemanticKITTI benchmarks that demonstrate the superiority of our proposed H3O.
comment: ICRA 2025
☆ Enhancing Alzheimer's Diagnosis: Leveraging Anatomical Landmarks in Graph Convolutional Neural Networks on Tetrahedral Meshes
Alzheimer's disease (AD) is a major neurodegenerative condition that affects millions around the world. As one of the main biomarkers in the AD diagnosis procedure, brain amyloid positivity is typically identified by positron emission tomography (PET), which is costly and invasive. Brain structural magnetic resonance imaging (sMRI) may provide a safer and more convenient solution for the AD diagnosis. Recent advances in geometric deep learning have facilitated sMRI analysis and early diagnosis of AD. However, determining AD pathology, such as brain amyloid deposition, in preclinical stage remains challenging, as less significant morphological changes can be observed. As a result, few AD classification models are generalizable to the brain amyloid positivity classification task. Blood-based biomarkers (BBBMs), on the other hand, have recently achieved remarkable success in predicting brain amyloid positivity and identifying individuals with high risk of being brain amyloid positive. However, individuals in medium risk group still require gold standard tests such as Amyloid PET for further evaluation. Inspired by the recent success of transformer architectures, we propose a geometric deep learning model based on transformer that is both scalable and robust to variations in input volumetric mesh size. Our work introduced a novel tokenization scheme for tetrahedral meshes, incorporating anatomical landmarks generated by a pre-trained Gaussian process model. Our model achieved superior classification performance in AD classification task. In addition, we showed that the model was also generalizable to the brain amyloid positivity prediction with individuals in the medium risk class, where BM alone cannot achieve a clear classification. Our work may enrich geometric deep learning research and improve AD diagnosis accuracy without using expensive and invasive PET scans.
☆ ISP-AD: A Large-Scale Real-World Dataset for Advancing Industrial Anomaly Detection with Synthetic and Real Defects
Automatic visual inspection using machine learning-based methods plays a key role in achieving zero-defect policies in industry. Research on anomaly detection approaches is constrained by the availability of datasets that represent complex defect appearances and imperfect imaging conditions, which are typical to industrial processes. Recent benchmarks indicate that most publicly available datasets are biased towards optimal imaging conditions, leading to an overestimation of the methods' applicability to real-world industrial scenarios. To address this gap, we introduce the Industrial Screen Printing Anomaly Detection dataset (ISP-AD). It presents challenging small and weakly contrasted surface defects embedded within structured patterns exhibiting high permitted design variability. To the best of our knowledge, it is the largest publicly available industrial dataset to date, including both synthetic and real defects collected directly from the factory floor. In addition to the evaluation of defect detection performance of recent unsupervised anomaly detection methods, experiments on a mixed supervised training approach, incorporating both synthesized and real defects, were conducted. Even small amounts of injected real defects prove beneficial for model generalization. Furthermore, starting from training on purely synthetic defects, emerging real defective samples can be efficiently integrated into subsequent scalable training. Research findings indicate that supervision by means of both synthetic and accumulated real defects can complement each other, meeting demanded industrial inspection requirements such as low false positive rates and high recall. The presented unsupervised and supervised dataset splits are designed to emphasize research on unsupervised, self-supervised, and supervised approaches, enhancing their applicability to industrial settings.
comment: 26 pages, 6 figures, this preprint has been submitted to the Journal of Intelligent Manufacturing
☆ Leveraging Large Language Models For Scalable Vector Graphics Processing: A Review
In recent years, rapid advances in computer vision have significantly improved the processing and generation of raster images. However, vector graphics, which is essential in digital design, due to its scalability and ease of editing, have been relatively understudied. Traditional vectorization techniques, which are often used in vector generation, suffer from long processing times and excessive output complexity, limiting their usability in practical applications. The advent of large language models (LLMs) has opened new possibilities for the generation, editing, and analysis of vector graphics, particularly in the SVG format, which is inherently text-based and well-suited for integration with LLMs. This paper provides a systematic review of existing LLM-based approaches for SVG processing, categorizing them into three main tasks: generation, editing, and understanding. We observe notable models such as IconShop, StrokeNUWA, and StarVector, highlighting their strengths and limitations. Furthermore, we analyze benchmark datasets designed for assessing SVG-related tasks, including SVGEditBench, VGBench, and SGP-Bench, and conduct a series of experiments to evaluate various LLMs in these domains. Our results demonstrate that for vector graphics reasoning-enhanced models outperform standard LLMs, particularly in generation and understanding tasks. Furthermore, our findings underscore the need to develop more diverse and richly annotated datasets to further improve LLM capabilities in vector graphics tasks.
☆ LVLM-Compress-Bench: Benchmarking the Broader Impact of Large Vision-Language Model Compression NAACL 2025
Despite recent efforts in understanding the compression impact on large language models (LLMs) in terms of their downstream task performance and trustworthiness on relatively simpler uni-modal benchmarks (for example, question answering, common sense reasoning), their detailed study on multi-modal Large Vision-Language Models (LVLMs) is yet to be unveiled. Towards mitigating this gap, we present LVLM-Compress-Bench, a framework to first thoroughly study the broad impact of compression on the generative performance of LVLMs with multi-modal input driven tasks. In specific, we consider two major classes of compression for autoregressive models, namely KV cache and weight compression, for the dynamically growing intermediate cache and static weights, respectively. We use four LVLM variants of the popular LLaVA framework to present our analysis via integrating various state-of-the-art KV and weight compression methods including uniform, outlier-reduced, and group quantization for the KV cache and weights. With this framework we demonstrate on ten different multi-modal datasets with different capabilities including recognition, knowledge, language generation, spatial awareness, visual reasoning, hallucination and visual illusion identification, toxicity, stereotypes and bias. In specific, our framework demonstrates the compression impact on both general and ethically critical metrics leveraging a combination of real world and synthetic datasets to encompass diverse societal intersectional attributes. Extensive experimental evaluations yield diverse and intriguing observations on the behavior of LVLMs at different quantization budget of KV and weights, in both maintaining and losing performance as compared to the baseline model with FP16 data format. Code will be open-sourced at https://github.com/opengear-project/LVLM-compress-bench.
comment: This work has been accepted to NAACL 2025
☆ HyDA: Hypernetworks for Test Time Domain Adaptation in Medical Imaging Analysis MICCAI 2025
Medical imaging datasets often vary due to differences in acquisition protocols, patient demographics, and imaging devices. These variations in data distribution, known as domain shift, present a significant challenge in adapting imaging analysis models for practical healthcare applications. Most current domain adaptation (DA) approaches aim either to align the distributions between the source and target domains or to learn an invariant feature space that generalizes well across all domains. However, both strategies require access to a sufficient number of examples, though not necessarily annotated, from the test domain during training. This limitation hinders the widespread deployment of models in clinical settings, where target domain data may only be accessible in real time. In this work, we introduce HyDA, a novel hypernetwork framework that leverages domain characteristics rather than suppressing them, enabling dynamic adaptation at inference time. Specifically, HyDA learns implicit domain representations and uses them to adjust model parameters on-the-fly, effectively interpolating to unseen domains. We validate HyDA on two clinically relevant applications - MRI brain age prediction and chest X-ray pathology classification - demonstrating its ability to generalize across tasks and modalities. Our code is available at TBD.
comment: submitted to MICCAI 2025
☆ Prediction of Frozen Region Growth in Kidney Cryoablation Intervention Using a 3D Flow-Matching Model MICCAI 2025
This study presents a 3D flow-matching model designed to predict the progression of the frozen region (iceball) during kidney cryoablation. Precise intraoperative guidance is critical in cryoablation to ensure complete tumor eradication while preserving adjacent healthy tissue. However, conventional methods, typically based on physics driven or diffusion based simulations, are computationally demanding and often struggle to represent complex anatomical structures accurately. To address these limitations, our approach leverages intraoperative CT imaging to inform the model. The proposed 3D flow matching model is trained to learn a continuous deformation field that maps early-stage CT scans to future predictions. This transformation not only estimates the volumetric expansion of the iceball but also generates corresponding segmentation masks, effectively capturing spatial and morphological changes over time. Quantitative analysis highlights the model robustness, demonstrating strong agreement between predictions and ground-truth segmentations. The model achieves an Intersection over Union (IoU) score of 0.61 and a Dice coefficient of 0.75. By integrating real time CT imaging with advanced deep learning techniques, this approach has the potential to enhance intraoperative guidance in kidney cryoablation, improving procedural outcomes and advancing the field of minimally invasive surgery.
comment: MICCAI 2025 submitted version
☆ Spectral Informed Mamba for Robust Point Cloud Processing
State space models have shown significant promise in Natural Language Processing (NLP) and, more recently, computer vision. This paper introduces a new methodology leveraging Mamba and Masked Autoencoder networks for point cloud data in both supervised and self-supervised learning. We propose three key contributions to enhance Mamba's capability in processing complex point cloud structures. First, we exploit the spectrum of a graph Laplacian to capture patch connectivity, defining an isometry-invariant traversal order that is robust to viewpoints and better captures shape manifolds than traditional 3D grid-based traversals. Second, we adapt segmentation via a recursive patch partitioning strategy informed by Laplacian spectral components, allowing finer integration and segment analysis. Third, we address token placement in Masked Autoencoder for Mamba by restoring tokens to their original positions, which preserves essential order and improves learning. Extensive experiments demonstrate the improvements of our approach in classification, segmentation, and few-shot tasks over state-of-the-art baselines.
☆ Metadata-free Georegistration of Ground and Airborne Imagery WACV 2025
Heterogeneous collections of ground and airborne imagery can readily be used to create high-quality 3D models and novel viewpoint renderings of the observed scene. Standard photogrammetry pipelines generate models in arbitrary coordinate systems, which is problematic for applications that require georegistered models. Even for applications that do not require georegistered models, georegistration is useful as a mechanism for aligning multiple disconnected models generated from non-overlapping data. The proposed method leverages satellite imagery, an associated digital surface model (DSM), and the novel view generation capabilities of modern 3D modeling techniques (e.g. neural radiance fields) to provide a robust method for georegistering airborne imagery, and a related technique for registering ground-based imagery to models created from airborne imagery. Experiments demonstrate successful georegistration of airborne and ground-based photogrammetric models across a variety of distinct sites. The proposed method does not require use of any metadata other than a satellite-based reference product and therefore has general applicability.
comment: WACV 2025 ULTRRA Workshop
♻ ☆ Reasoning to Attend: Try to Understand How Token Works CVPR 2025
Current Large Multimodal Models (LMMs) empowered visual grounding typically rely on $\texttt{}$ tokens as a text prompt to jointly optimize the vision-language model (e.g., LLaVA) and the downstream task-specific model (e.g., SAM). However, we observe that little research has looked into how it works.In this work, we first visualize the similarity maps, which are obtained by computing the semantic similarity between the $\texttt{}$ token and the image token embeddings derived from the last hidden layer in both the LLaVA encoder and SAM decoder. Intriguingly, we have found that a striking consistency holds in terms of activation responses in the similarity map, which reveals that what the $\texttt{}$ token contributes to is semantic similarity within image-text pairs. Specifically, the $\texttt{}$ token, a placeholder expanded in text vocabulary, extensively queries among individual tokenized image patches to match the semantics of an object from text to the paired image, while the Large Language Models (LLMs) are being fine-tuned. Upon the above findings, we present READ, which facilitates LMMs' resilient $\textbf{REA}$soning capability of where to atten$\textbf{D}$ under the guidance of highly activated points borrowed from similarity maps. Remarkably, READ features an intuitive design, Similarity as Points module (SasP), which can be seamlessly applied to $\texttt{}$-like paradigms in a plug-and-play fashion. Also, extensive experiments have been conducted on ReasonSeg and RefCOCO(+/g) datasets. To validate whether READ suffers from catastrophic forgetting of previous skills after fine-tuning, we further assess its generation ability on an augmented FP-RefCOCO(+/g) dataset. All codes and models are publicly available at https://github.com/rui-qian/READ.
comment: This work has been accepted to CVPR 2025, please refer to https://github.com/rui-qian/READ
♻ ☆ Towards Visual Discrimination and Reasoning of Real-World Physical Dynamics: Physics-Grounded Anomaly Detection CVPR 2025
Humans detect real-world object anomalies by perceiving, interacting, and reasoning based on object-conditioned physical knowledge. The long-term goal of Industrial Anomaly Detection (IAD) is to enable machines to autonomously replicate this skill. However, current IAD algorithms are largely developed and tested on static, semantically simple datasets, which diverge from real-world scenarios where physical understanding and reasoning are essential. To bridge this gap, we introduce the Physics Anomaly Detection (Phys-AD) dataset, the first large-scale, real-world, physics-grounded video dataset for industrial anomaly detection. Collected using a real robot arm and motor, Phys-AD provides a diverse set of dynamic, semantically rich scenarios. The dataset includes more than 6400 videos across 22 real-world object categories, interacting with robot arms and motors, and exhibits 47 types of anomalies. Anomaly detection in Phys-AD requires visual reasoning, combining both physical knowledge and video content to determine object abnormality. We benchmark state-of-the-art anomaly detection methods under three settings: unsupervised AD, weakly-supervised AD, and video-understanding AD, highlighting their limitations in handling physics-grounded anomalies. Additionally, we introduce the Physics Anomaly Explanation (PAEval) metric, designed to assess the ability of visual-language foundation models to not only detect anomalies but also provide accurate explanations for their underlying physical causes. Our dataset and benchmark will be publicly available.
comment: Accepted by CVPR 2025
♻ ☆ DTU-Net: A Multi-Scale Dilated Transformer Network for Nonlinear Hyperspectral Unmixing
Transformers have shown significant success in hyperspectral unmixing (HU). However, challenges remain. While multi-scale and long-range spatial correlations are essential in unmixing tasks, current Transformer-based unmixing networks, built on Vision Transformer (ViT) or Swin-Transformer, struggle to capture them effectively. Additionally, current Transformer-based unmixing networks rely on the linear mixing model, which lacks the flexibility to accommodate scenarios where nonlinear effects are significant. To address these limitations, we propose a multi-scale Dilated Transformer-based unmixing network for nonlinear HU (DTU-Net). The encoder employs two branches. The first one performs multi-scale spatial feature extraction using Multi-Scale Dilated Attention (MSDA) in the Dilated Transformer, which varies dilation rates across attention heads to capture long-range and multi-scale spatial correlations. The second one performs spectral feature extraction utilizing 3D-CNNs with channel attention. The outputs from both branches are then fused to integrate multi-scale spatial and spectral information, which is subsequently transformed to estimate the abundances. The decoder is designed to accommodate both linear and nonlinear mixing scenarios. Its interpretability is enhanced by explicitly modeling the relationships between endmembers, abundances, and nonlinear coefficients in accordance with the polynomial post-nonlinear mixing model (PPNMM). Experiments on synthetic and real datasets validate the effectiveness of the proposed DTU-Net compared to PPNMM-derived methods and several advanced unmixing networks.
♻ ☆ Explaining Caption-Image Interactions in CLIP models with Second-Order Attributions
Dual encoder architectures like CLIP models map two types of inputs into a shared embedding space and predict similarities between them. Despite their success, it is, however, not understood how these models compare their two inputs. Common first-order feature-attribution methods can only provide limited insights into dual-encoders since their predictions depend on feature-interactions rather than on individual features. In this paper, we first derive a second-order method enabling the attribution of predictions by any differentiable dual encoder onto feature-interactions between its inputs. Second, we apply our method to CLIP models and show that they learn fine-grained correspondences between parts of captions and regions in images. They match objects across input modes also account for mismatches. This visual-linguistic grounding ability, however, varies heavily between object classes and exhibits pronounced out-of-domain effects. We can identify individual errors as well as systematic failure categories including object coverage, unusual scenes and correlated contexts.
♻ ☆ Optimizing for the Shortest Path in Denoising Diffusion Model CVPR 2025
In this research, we propose a novel denoising diffusion model based on shortest-path modeling that optimizes residual propagation to enhance both denoising efficiency and quality. Drawing on Denoising Diffusion Implicit Models (DDIM) and insights from graph theory, our model, termed the Shortest Path Diffusion Model (ShortDF), treats the denoising process as a shortest-path problem aimed at minimizing reconstruction error. By optimizing the initial residuals, we improve the efficiency of the reverse diffusion process and the quality of the generated samples. Extensive experiments on multiple standard benchmarks demonstrate that ShortDF significantly reduces diffusion time (or steps) while enhancing the visual fidelity of generated samples compared to prior arts. This work, we suppose, paves the way for interactive diffusion-based applications and establishes a foundation for rapid data generation. Code is available at https://github.com/UnicomAI/ShortDF
comment: Accepet by CVPR 2025 (10 pages, 6 figures)
♻ ☆ DSPNet: Dual-vision Scene Perception for Robust 3D Question Answering CVPR 2025
3D Question Answering (3D QA) requires the model to comprehensively understand its situated 3D scene described by the text, then reason about its surrounding environment and answer a question under that situation. However, existing methods usually rely on global scene perception from pure 3D point clouds and overlook the importance of rich local texture details from multi-view images. Moreover, due to the inherent noise in camera poses and complex occlusions, there exists significant feature degradation and reduced feature robustness problems when aligning 3D point cloud with multi-view images. In this paper, we propose a Dual-vision Scene Perception Network (DSPNet), to comprehensively integrate multi-view and point cloud features to improve robustness in 3D QA. Our Text-guided Multi-view Fusion (TGMF) module prioritizes image views that closely match the semantic content of the text. To adaptively fuse back-projected multi-view images with point cloud features, we design the Adaptive Dual-vision Perception (ADVP) module, enhancing 3D scene comprehension. Additionally, our Multimodal Context-guided Reasoning (MCGR) module facilitates robust reasoning by integrating contextual information across visual and linguistic modalities. Experimental results on SQA3D and ScanQA datasets demonstrate the superiority of our DSPNet. Codes will be available at https://github.com/LZ-CH/DSPNet.
comment: Accepted by CVPR 2025
♻ ☆ Detecting Systematic Weaknesses in Vision Models along Predefined Human-Understandable Dimensions
Slice discovery methods (SDMs) are prominent algorithms for finding systematic weaknesses in DNNs. They identify top-k semantically coherent slices/subsets of data where a DNN-under-test has low performance. For being directly useful, slices should be aligned with human-understandable and relevant dimensions, which, for example, are defined by safety and domain experts as part of the operational design domain (ODD). While SDMs can be applied effectively on structured data, their application on image data is complicated by the lack of semantic metadata. To address these issues, we present an algorithm that combines foundation models for zero-shot image classification to generate semantic metadata with methods for combinatorial search to find systematic weaknesses in images. In contrast to existing approaches, ours identifies weak slices that are in line with pre-defined human-understandable dimensions. As the algorithm includes foundation models, its intermediate and final results may not always be exact. Therefore, we include an approach to address the impact of noisy metadata. We validate our algorithm on both synthetic and real-world datasets, demonstrating its ability to recover human-understandable systematic weaknesses. Furthermore, using our approach, we identify systematic weaknesses of multiple pre-trained and publicly available state-of-the-art computer vision DNNs.
♻ ☆ ZeroBench: An Impossible Visual Benchmark for Contemporary Large Multimodal Models
Large Multimodal Models (LMMs) exhibit major shortfalls when interpreting images and, by some measures, have poorer spatial cognition than small children or animals. Despite this, they attain high scores on many popular visual benchmarks, with headroom rapidly eroded by an ongoing surge of model progress. To address this, there is a pressing need for difficult benchmarks that remain relevant for longer. We take this idea to its limit by introducing ZeroBench-a lightweight visual reasoning benchmark that is entirely impossible for contemporary frontier LMMs. Our benchmark consists of 100 manually curated questions and 334 less difficult subquestions. We evaluate 20 LMMs on ZeroBench, all of which score 0.0%, and rigorously analyse the errors. To encourage progress in visual understanding, we publicly release ZeroBench.
comment: 20 pages, 13 figures
♻ ☆ Back Home: A Machine Learning Approach to Seashell Classification and Ecosystem Restoration
In Costa Rica, an average of 5 tons of seashells are extracted from ecosystems annually. Confiscated seashells, cannot be returned to their ecosystems due to the lack of origin recognition. To address this issue, we developed a convolutional neural network (CNN) specifically for seashell identification. We built a dataset from scratch, consisting of approximately 19000 images from the Pacific and Caribbean coasts. Using this dataset, the model achieved a classification accuracy exceeding 85%. The model has been integrated into a user-friendly application, which has classified over 36,000 seashells to date, delivering real-time results within 3 seconds per image. To further enhance the system's accuracy, an anomaly detection mechanism was incorporated to filter out irrelevant or anomalous inputs, ensuring only valid seashell images are processed.
♻ ☆ A Simple and Effective Reinforcement Learning Method for Text-to-Image Diffusion Fine-tuning
Reinforcement learning (RL)-based fine-tuning has emerged as a powerful approach for aligning diffusion models with black-box objectives. Proximal policy optimization (PPO) is the most popular choice of method for policy optimization. While effective in terms of performance, PPO is highly sensitive to hyper-parameters and involves substantial computational overhead. REINFORCE, on the other hand, mitigates some computational complexities such as high memory overhead and sensitive hyper-parameter tuning, but has suboptimal performance due to high-variance and sample inefficiency. While the variance of the REINFORCE can be reduced by sampling multiple actions per input prompt and using a baseline correction term, it still suffers from sample inefficiency. To address these challenges, we systematically analyze the efficiency-effectiveness trade-off between REINFORCE and PPO, and propose leave-one-out PPO (LOOP), a novel RL for diffusion fine-tuning method. LOOP combines variance reduction techniques from REINFORCE, such as sampling multiple actions per input prompt and a baseline correction term, with the robustness and sample efficiency of PPO via clipping and importance sampling. Our results demonstrate that LOOP effectively improves diffusion models on various black-box objectives, and achieves a better balance between computational efficiency and performance.
♻ ☆ A Survey of Deep Learning-based Radiology Report Generation Using Multimodal Data
Automatic radiology report generation can alleviate the workload for physicians and minimize regional disparities in medical resources, therefore becoming an important topic in the medical image analysis field. It is a challenging task, as the computational model needs to mimic physicians to obtain information from multi-modal input data (i.e., medical images, clinical information, medical knowledge, etc.), and produce comprehensive and accurate reports. Recently, numerous works have emerged to address this issue using deep-learning-based methods, such as transformers, contrastive learning, and knowledge-base construction. This survey summarizes the key techniques developed in the most recent works and proposes a general workflow for deep-learning-based report generation with five main components, including multi-modality data acquisition, data preparation, feature learning, feature fusion and interaction, and report generation. The state-of-the-art methods for each of these components are highlighted. Additionally, we summarize the latest developments in large model-based methods and model explainability, along with public datasets, evaluation methods, current challenges, and future directions in this field. We have also conducted a quantitative comparison between different methods in the same experimental setting. This is the most up-to-date survey that focuses on multi-modality inputs and data fusion for radiology report generation. The aim is to provide comprehensive and rich information for researchers interested in automatic clinical report generation and medical image analysis, especially when using multimodal inputs, and to assist them in developing new algorithms to advance the field.
♻ ☆ LLM-wrapper: Black-Box Semantic-Aware Adaptation of Vision-Language Models for Referring Expression Comprehension ICLR 2025
Vision Language Models (VLMs) have demonstrated remarkable capabilities in various open-vocabulary tasks, yet their zero-shot performance lags behind task-specific fine-tuned models, particularly in complex tasks like Referring Expression Comprehension (REC). Fine-tuning usually requires 'white-box' access to the model's architecture and weights, which is not always feasible due to proprietary or privacy concerns. In this work, we propose LLM-wrapper, a method for 'black-box' adaptation of VLMs for the REC task using Large Language Models (LLMs). LLM-wrapper capitalizes on the reasoning abilities of LLMs, improved with a light fine-tuning, to select the most relevant bounding box matching the referring expression, from candidates generated by a zero-shot black-box VLM. Our approach offers several advantages: it enables the adaptation of closed-source models without needing access to their internal workings, it is versatile as it works with any VLM, it transfers to new VLMs and datasets, and it allows for the adaptation of an ensemble of VLMs. We evaluate LLM-wrapper on multiple datasets using different VLMs and LLMs, demonstrating significant performance improvements and highlighting the versatility of our method. While LLM-wrapper is not meant to directly compete with standard white-box fine-tuning, it offers a practical and effective alternative for black-box VLM adaptation. Code and checkpoints are available at https://github.com/valeoai/LLM_wrapper .
comment: LLM-wrapper (v3) is published as a conference paper at ICLR 2025. (v1 was presented at EVAL-FoMo workshop, ECCV 2024.)
♻ ☆ Human-Feedback Efficient Reinforcement Learning for Online Diffusion Model Finetuning ICLR
Controllable generation through Stable Diffusion (SD) fine-tuning aims to improve fidelity, safety, and alignment with human guidance. Existing reinforcement learning from human feedback methods usually rely on predefined heuristic reward functions or pretrained reward models built on large-scale datasets, limiting their applicability to scenarios where collecting such data is costly or difficult. To effectively and efficiently utilize human feedback, we develop a framework, HERO, which leverages online human feedback collected on the fly during model learning. Specifically, HERO features two key mechanisms: (1) Feedback-Aligned Representation Learning, an online training method that captures human feedback and provides informative learning signals for fine-tuning, and (2) Feedback-Guided Image Generation, which involves generating images from SD's refined initialization samples, enabling faster convergence towards the evaluator's intent. We demonstrate that HERO is 4x more efficient in online feedback for body part anomaly correction compared to the best existing method. Additionally, experiments show that HERO can effectively handle tasks like reasoning, counting, personalization, and reducing NSFW content with only 0.5K online feedback.
comment: Published in International Conference on Learning Representations (ICLR) 2025
♻ ☆ BHViT: Binarized Hybrid Vision Transformer CVPR2025
Model binarization has made significant progress in enabling real-time and energy-efficient computation for convolutional neural networks (CNN), offering a potential solution to the deployment challenges faced by Vision Transformers (ViTs) on edge devices. However, due to the structural differences between CNN and Transformer architectures, simply applying binary CNN strategies to the ViT models will lead to a significant performance drop. To tackle this challenge, we propose BHViT, a binarization-friendly hybrid ViT architecture and its full binarization model with the guidance of three important observations. Initially, BHViT utilizes the local information interaction and hierarchical feature aggregation technique from coarse to fine levels to address redundant computations stemming from excessive tokens. Then, a novel module based on shift operations is proposed to enhance the performance of the binary Multilayer Perceptron (MLP) module without significantly increasing computational overhead. In addition, an innovative attention matrix binarization method based on quantization decomposition is proposed to evaluate the token's importance in the binarized attention matrix. Finally, we propose a regularization loss to address the inadequate optimization caused by the incompatibility between the weight oscillation in the binary layers and the Adam Optimizer. Extensive experimental results demonstrate that our proposed algorithm achieves SOTA performance among binary ViT methods.
comment: Accepted by CVPR2025
♻ ☆ Self-supervised pre-training with diffusion model for few-shot landmark detection in x-ray images WACV 2025
Deep neural networks have been extensively applied in the medical domain for various tasks, including image classification, segmentation, and landmark detection. However, their application is often hindered by data scarcity, both in terms of available annotations and images. This study introduces a novel application of denoising diffusion probabilistic models (DDPMs) to the landmark detection task, specifically addressing the challenge of limited annotated data in x-ray imaging. Our key innovation lies in leveraging DDPMs for self-supervised pre-training in landmark detection, a previously unexplored approach in this domain. This method enables accurate landmark detection with minimal annotated training data (as few as 50 images), surpassing both ImageNet supervised pre-training and traditional self-supervised techniques across three popular x-ray benchmark datasets. To our knowledge, this work represents the first application of diffusion models for self-supervised learning in landmark detection, which may offer a valuable pre-training approach in few-shot regimes, for mitigating data scarcity.
comment: Accepted at WACV 2025
♻ ☆ Enhancing Multimodal Medical Image Classification using Cross-Graph Modal Contrastive Learning
The classification of medical images is a pivotal aspect of disease diagnosis, often enhanced by deep learning techniques. However, traditional approaches typically focus on unimodal medical image data, neglecting the integration of diverse non-image patient data. This paper proposes a novel Cross-Graph Modal Contrastive Learning (CGMCL) framework for multimodal structured data from different data domains to improve medical image classification. The model effectively integrates both image and non-image data by constructing cross-modality graphs and leveraging contrastive learning to align multimodal features in a shared latent space. An inter-modality feature scaling module further optimizes the representation learning process by reducing the gap between heterogeneous modalities. The proposed approach is evaluated on two datasets: a Parkinson's disease (PD) dataset and a public melanoma dataset. Results demonstrate that CGMCL outperforms conventional unimodal methods in accuracy, interpretability, and early disease prediction. Additionally, the method shows superior performance in multi-class melanoma classification. The CGMCL framework provides valuable insights into medical image classification while offering improved disease interpretability and predictive capabilities.
♻ ☆ LION-FS: Fast & Slow Video-Language Thinker as Online Video Assistant CVPR 2025
First-person video assistants are highly anticipated to enhance our daily lives through online video dialogue. However, existing online video assistants often sacrifice assistant efficacy for real-time efficiency by processing low-frame-rate videos with coarse-grained visual features.To overcome the trade-off between efficacy and efficiency, we propose "Fast & Slow Video-Language Thinker" as an onLIne videO assistaNt, LION-FS, achieving real-time, proactive, temporally accurate, and contextually precise responses. LION-FS adopts a two-stage optimization strategy: 1)Fast Path: Routing-Based Response Determination evaluates frame-by-frame whether an immediate response is necessary. To enhance response determination accuracy and handle higher frame-rate inputs efficiently, we employ Token Aggregation Routing to dynamically fuse spatiotemporal features without increasing token numbers, while utilizing Token Dropping Routing to eliminate redundant features. 2)Slow Path: Multi-granularity Keyframe Augmentation optimizes keyframes during response generation. To provide comprehensive and detailed responses beyond atomic actions constrained by training data, fine-grained spatial features and human-environment interaction features are extracted through multi-granular pooling. These features are further integrated into a meticulously designed multimodal Thinking Template to guide more precise response generation. Comprehensive evaluations on online video tasks demonstrate that LION-FS achieves state-of-the-art efficacy and efficiency.
comment: Accept to CVPR 2025, Project page: https://github.com/JiuTian-VL/LION-FS
♻ ☆ X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability
Despite the rapid development of safety alignment techniques for LLMs, defending against multi-turn jailbreaks is still a challenging task. In this paper, we conduct a comprehensive comparison, revealing that some existing defense methods can improve the robustness of LLMs against multi-turn jailbreaks but compromise usability, i.e., reducing general capabilities or causing the over-refusal problem. From the perspective of mechanism interpretability of LLMs, we discover that these methods fail to establish a boundary that exactly distinguishes safe and harmful feature representations. Therefore, boundary-safe representations close to harmful representations are inevitably disrupted, leading to a decline in usability. To address this issue, we propose X-Boundary to push harmful representations away from boundary-safe representations and obtain an exact distinction boundary. In this way, harmful representations can be precisely erased without disrupting safe ones. Experimental results show that X-Boundary achieves state-of-the-art defense performance against multi-turn jailbreaks, while reducing the over-refusal rate by about 20% and maintaining nearly complete general capability. Furthermore, we theoretically prove and empirically verify that X-Boundary can accelerate the convergence process during training. Please see our code at: https://github.com/AI45Lab/X-Boundary.
♻ ☆ GSPR: Multimodal Place Recognition Using 3D Gaussian Splatting for Autonomous Driving
Place recognition is a crucial component that enables autonomous vehicles to obtain localization results in GPS-denied environments. In recent years, multimodal place recognition methods have gained increasing attention. They overcome the weaknesses of unimodal sensor systems by leveraging complementary information from different modalities. However, most existing methods explore cross-modality correlations through feature-level or descriptor-level fusion, suffering from a lack of interpretability. Conversely, the recently proposed 3D Gaussian Splatting provides a new perspective on multimodal fusion by harmonizing different modalities into an explicit scene representation. In this paper, we propose a 3D Gaussian Splatting-based multimodal place recognition network dubbed GSPR. It explicitly combines multi-view RGB images and LiDAR point clouds into a spatio-temporally unified scene representation with the proposed Multimodal Gaussian Splatting. A network composed of 3D graph convolution and transformer is designed to extract spatio-temporal features and global descriptors from the Gaussian scenes for place recognition. Extensive evaluations on three datasets demonstrate that our method can effectively leverage complementary strengths of both multi-view cameras and LiDAR, achieving SOTA place recognition performance while maintaining solid generalization ability. Our open-source code will be released at https://github.com/QiZS-BIT/GSPR.
comment: 8 pages, 6 figures
Efficient Diversity-Preserving Diffusion Alignment via Gradient-Informed GFlowNets ICLR 2025
While one commonly trains large diffusion models by collecting datasets on target downstream tasks, it is often desired to align and finetune pretrained diffusion models with some reward functions that are either designed by experts or learned from small-scale datasets. Existing post-training methods for reward finetuning of diffusion models typically suffer from lack of diversity in generated samples, lack of prior preservation, and/or slow convergence in finetuning. Inspired by recent successes in generative flow networks (GFlowNets), a class of probabilistic models that sample with the unnormalized density of a reward function, we propose a novel GFlowNet method dubbed Nabla-GFlowNet (abbreviated as \methodname), the first GFlowNet method that leverages the rich signal in reward gradients, together with an objective called \graddb plus its variant \resgraddb designed for prior-preserving diffusion finetuning. We show that our proposed method achieves fast yet diversity- and prior-preserving finetuning of Stable Diffusion, a large-scale text-conditioned image diffusion model, on different realistic reward functions.
comment: Technical Report (35 pages, 31 figures), Accepted at ICLR 2025
♻ ☆ MMGDreamer: Mixed-Modality Graph for Geometry-Controllable 3D Indoor Scene Generation AAAI 2025
Controllable 3D scene generation has extensive applications in virtual reality and interior design, where the generated scenes should exhibit high levels of realism and controllability in terms of geometry. Scene graphs provide a suitable data representation that facilitates these applications. However, current graph-based methods for scene generation are constrained to text-based inputs and exhibit insufficient adaptability to flexible user inputs, hindering the ability to precisely control object geometry. To address this issue, we propose MMGDreamer, a dual-branch diffusion model for scene generation that incorporates a novel Mixed-Modality Graph, visual enhancement module, and relation predictor. The mixed-modality graph allows object nodes to integrate textual and visual modalities, with optional relationships between nodes. It enhances adaptability to flexible user inputs and enables meticulous control over the geometry of objects in the generated scenes. The visual enhancement module enriches the visual fidelity of text-only nodes by constructing visual representations using text embeddings. Furthermore, our relation predictor leverages node representations to infer absent relationships between nodes, resulting in more coherent scene layouts. Extensive experimental results demonstrate that MMGDreamer exhibits superior control of object geometry, achieving state-of-the-art scene generation performance. Project page: https://yangzhifeio.github.io/project/MMGDreamer.
comment: Accepted by AAAI 2025 Main Track
♻ ☆ FSPGD: Rethinking Black-box Attacks on Semantic Segmentation
Transferability, the ability of adversarial examples crafted for one model to deceive other models, is crucial for black-box attacks. Despite advancements in attack methods for semantic segmentation, transferability remains limited, reducing their effectiveness in real-world applications. To address this, we introduce the Feature Similarity Projected Gradient Descent (FSPGD) attack, a novel black-box approach that enhances both attack performance and transferability. Unlike conventional segmentation attacks that rely on output predictions for gradient calculation, FSPGD computes gradients from intermediate layer features. Specifically, our method introduces a loss function that targets local information by comparing features between clean images and adversarial examples, while also disrupting contextual information by accounting for spatial relationships between objects. Experiments on Pascal VOC 2012 and Cityscapes datasets demonstrate that FSPGD achieves superior transferability and attack performance, establishing a new state-of-the-art benchmark. Code is available at https://github.com/KU-AIVS/FSPGD.
♻ ☆ UniMLVG: Unified Framework for Multi-view Long Video Generation with Comprehensive Control Capabilities for Autonomous Driving
The creation of diverse and realistic driving scenarios has become essential to enhance perception and planning capabilities of the autonomous driving system. However, generating long-duration, surround-view consistent driving videos remains a significant challenge. To address this, we present UniMLVG, a unified framework designed to generate extended street multi-perspective videos under precise control. By integrating single- and multi-view driving videos into the training data, our approach updates a DiT-based diffusion model equipped with cross-frame and cross-view modules across three stages with multi training objectives, substantially boosting the diversity and quality of generated visual content. Importantly, we propose an innovative explicit viewpoint modeling approach for multi-view video generation to effectively improve motion transition consistency. Capable of handling various input reference formats (e.g., text, images, or video), our UniMLVG generates high-quality multi-view videos according to the corresponding condition constraints such as 3D bounding boxes or frame-level text descriptions. Compared to the best models with similar capabilities, our framework achieves improvements of 48.2% in FID and 35.2% in FVD.
♻ ☆ Mocap-2-to-3: Lifting 2D Diffusion-Based Pretrained Models for 3D Motion Capture
Recovering absolute poses in the world coordinate system from monocular views presents significant challenges. Two primary issues arise in this context. Firstly, existing methods rely on 3D motion data for training, which requires collection in limited environments. Acquiring such 3D labels for new actions in a timely manner is impractical, severely restricting the model's generalization capabilities. In contrast, 2D poses are far more accessible and easier to obtain. Secondly, estimating a person's absolute position in metric space from a single viewpoint is inherently more complex. To address these challenges, we introduce Mocap-2-to-3, a novel framework that decomposes intricate 3D motions into 2D poses, leveraging 2D data to enhance 3D motion reconstruction in diverse scenarios and accurately predict absolute positions in the world coordinate system. We initially pretrain a single-view diffusion model with extensive 2D data, followed by fine-tuning a multi-view diffusion model for view consistency using publicly available 3D data. This strategy facilitates the effective use of large-scale 2D data. Additionally, we propose an innovative human motion representation that decouples local actions from global movements and encodes geometric priors of the ground, ensuring the generative model learns accurate motion priors from 2D data. During inference, this allows for the gradual recovery of global movements, resulting in more plausible positioning. We evaluate our model's performance on real-world datasets, demonstrating superior accuracy in motion and absolute human positioning compared to state-of-the-art methods, along with enhanced generalization and scalability. Our code will be made publicly available.
♻ ☆ MobileViM: A Light-weight and Dimension-independent Vision Mamba for 3D Medical Image Analysis
Efficient evaluation of three-dimensional (3D) medical images is crucial for diagnostic and therapeutic practices in healthcare. Recent years have seen a substantial uptake in applying deep learning and computer vision to analyse and interpret medical images. Traditional approaches, such as convolutional neural networks (CNNs) and vision transformers (ViTs), face significant computational challenges, prompting the need for architectural advancements. Recent efforts have led to the introduction of novel architectures like the ``Mamba'' model as alternative solutions to traditional CNNs or ViTs. The Mamba model excels in the linear processing of one-dimensional data with low computational demands. However, Mamba's potential for 3D medical image analysis remains underexplored and could face significant computational challenges as the dimension increases. This manuscript presents MobileViM, a streamlined architecture for efficient segmentation of 3D medical images. In the MobileViM network, we invent a new dimension-independent mechanism and a dual-direction traversing approach to incorporate with a vision-Mamba-based framework. MobileViM also features a cross-scale bridging technique to improve efficiency and accuracy across various medical imaging modalities. With these enhancements, MobileViM achieves segmentation speeds exceeding 90 frames per second (FPS) on a single graphics processing unit (i.e., NVIDIA RTX 4090). This performance is over 24 FPS faster than the state-of-the-art deep learning models for processing 3D images with the same computational resources. In addition, experimental evaluations demonstrate that MobileViM delivers superior performance, with Dice similarity scores reaching 92.72%, 86.69%, 80.46%, and 77.43% for PENGWIN, BraTS2024, ATLAS, and Toothfairy2 datasets, respectively, which significantly surpasses existing models.
comment: The corresponding author disagrees with the manuscript submitted to arXiv
♻ ☆ FRNet: Frustum-Range Networks for Scalable LiDAR Segmentation
LiDAR segmentation has become a crucial component of advanced autonomous driving systems. Recent range-view LiDAR segmentation approaches show promise for real-time processing. However, they inevitably suffer from corrupted contextual information and rely heavily on post-processing techniques for prediction refinement. In this work, we propose FRNet, a simple yet powerful method aimed at restoring the contextual information of range image pixels using corresponding frustum LiDAR points. First, a frustum feature encoder module is used to extract per-point features within the frustum region, which preserves scene consistency and is critical for point-level predictions. Next, a frustum-point fusion module is introduced to update per-point features hierarchically, enabling each point to extract more surrounding information through the frustum features. Finally, a head fusion module is used to fuse features at different levels for final semantic predictions. Extensive experiments conducted on four popular LiDAR segmentation benchmarks under various task setups demonstrate the superiority of FRNet. Notably, FRNet achieves 73.3% and 82.5% mIoU scores on the testing sets of SemanticKITTI and nuScenes. While achieving competitive performance, FRNet operates 5 times faster than state-of-the-art approaches. Such high efficiency opens up new possibilities for more scalable LiDAR segmentation. The code has been made publicly available at https://github.com/Xiangxu-0103/FRNet.
comment: TIP 2025; 18 pages, 11 figures, 14 tables; Code at https://github.com/Xiangxu-0103/FRNet
♻ ☆ A Dataset and Benchmark for Shape Completion of Fruits for Agricultural Robotics
As the world population is expected to reach 10 billion by 2050, our agricultural production system needs to double its productivity despite a decline of human workforce in the agricultural sector. Autonomous robotic systems are one promising pathway to increase productivity by taking over labor-intensive manual tasks like fruit picking. To be effective, such systems need to monitor and interact with plants and fruits precisely, which is challenging due to the cluttered nature of agricultural environments causing, for example, strong occlusions. Thus, being able to estimate the complete 3D shapes of objects in presence of occlusions is crucial for automating operations such as fruit harvesting. In this paper, we propose the first publicly available 3D shape completion dataset for agricultural vision systems. We provide an RGB-D dataset for estimating the 3D shape of fruits. Specifically, our dataset contains RGB-D frames of single sweet peppers in lab conditions but also in a commercial greenhouse. For each fruit, we additionally collected high-precision point clouds that we use as ground truth. For acquiring the ground truth shape, we developed a measuring process that allows us to record data of real sweet pepper plants, both in the lab and in the greenhouse with high precision, and determine the shape of the sensed fruits. We release our dataset, consisting of almost 7,000 RGB-D frames belonging to more than 100 different fruits. We provide segmented RGB-D frames, with camera intrinsics to easily obtain colored point clouds, together with the corresponding high-precision, occlusion-free point clouds obtained with a high-precision laser scanner. We additionally enable evaluation of shape completion approaches on a hidden test set through a public challenge on a benchmark server.
♻ ☆ Towards Effective and Sparse Adversarial Attack on Spiking Neural Networks via Breaking Invisible Surrogate Gradients CVPR 2025
Spiking neural networks (SNNs) have shown their competence in handling spatial-temporal event-based data with low energy consumption. Similar to conventional artificial neural networks (ANNs), SNNs are also vulnerable to gradient-based adversarial attacks, wherein gradients are calculated by spatial-temporal back-propagation (STBP) and surrogate gradients (SGs). However, the SGs may be invisible for an inference-only model as they do not influence the inference results, and current gradient-based attacks are ineffective for binary dynamic images captured by the dynamic vision sensor (DVS). While some approaches addressed the issue of invisible SGs through universal SGs, their SGs lack a correlation with the victim model, resulting in sub-optimal performance. Moreover, the imperceptibility of existing SNN-based binary attacks is still insufficient. In this paper, we introduce an innovative potential-dependent surrogate gradient (PDSG) method to establish a robust connection between the SG and the model, thereby enhancing the adaptability of adversarial attacks across various models with invisible SGs. Additionally, we propose the sparse dynamic attack (SDA) to effectively attack binary dynamic images. Utilizing a generation-reduction paradigm, SDA can fully optimize the sparsity of adversarial perturbations. Experimental results demonstrate that our PDSG and SDA outperform state-of-the-art SNN-based attacks across various models and datasets. Specifically, our PDSG achieves 100% attack success rate on ImageNet, and our SDA obtains 82% attack success rate by modifying only 0.24% of the pixels on CIFAR10DVS. The code is available at https://github.com/ryime/PDSG-SDA .
comment: Accepted by CVPR 2025
♻ ☆ Continual Learning in 3D Point Clouds: Employing Spectral Techniques for Exemplar Selection WACV 2025
We introduce a novel framework for Continual Learning in 3D object classification. Our approach, CL3D, is based on the selection of prototypes from each class using spectral clustering. For non-Euclidean data such as point clouds, spectral clustering can be employed as long as one can define a distance measure between pairs of samples. Choosing the appropriate distance measure enables us to leverage 3D geometric characteristics to identify representative prototypes for each class. We explore the effectiveness of clustering in the input space (3D points), local feature space (1024-dimensional points), and global feature space. We conduct experiments on the ModelNet40, ShapeNet, and ScanNet datasets, achieving state-of-the-art accuracy exclusively through the use of input space features. By leveraging the combined input, local, and global features, we have improved the state-of-the-art on ModelNet and ShapeNet, utilizing nearly half the memory used by competing approaches. For the challenging ScanNet dataset, our method enhances accuracy by 4.1% while consuming just 28% of the memory used by our competitors, demonstrating the scalability of our approach.
comment: Accepted to WACV 2025, Tucson, Arizona, USA
♻ ☆ Implantable Adaptive Cells: A Novel Enhancement for Pre-Trained U-Nets in Medical Image Segmentation
This paper introduces a novel approach to enhance the performance of pre-trained neural networks in medical image segmentation using gradient-based Neural Architecture Search (NAS) methods. We present the concept of Implantable Adaptive Cell (IAC), small modules identified through Partially-Connected DARTS based approach, designed to be injected into the skip connections of an existing and already trained U-shaped model. Unlike traditional NAS methods, our approach refines existing architectures without full retraining. Experiments on four medical datasets with MRI and CT images show consistent accuracy improvements on various U-Net configurations, with segmentation accuracy gain by approximately 5 percentage points across all validation datasets, with improvements reaching up to 11\%pt in the best-performing cases. The findings of this study not only offer a cost-effective alternative to the complete overhaul of complex models for performance upgrades but also indicate the potential applicability of our method to other architectures and problem domains.
♻ ☆ Structured Preference Optimization for Vision-Language Long-Horizon Task Planning
Existing methods for vision-language task planning excel in short-horizon tasks but often fall short in complex, long-horizon planning within dynamic environments. These challenges primarily arise from the difficulty of effectively training models to produce high-quality reasoning processes for long-horizon tasks. To address this, we propose Structured Preference Optimization (SPO), which aims to enhance reasoning and action selection in long-horizon task planning through structured preference evaluation and optimized training strategies. Specifically, SPO introduces: 1) Preference-Based Scoring and Optimization, which systematically evaluates reasoning chains based on task relevance, visual grounding, and historical consistency; and 2) Curriculum-Guided Training, where the model progressively adapts from simple to complex tasks, improving its generalization ability in long-horizon scenarios and enhancing reasoning robustness. To advance research in vision-language long-horizon task planning, we introduce ExtendaBench, a comprehensive benchmark covering 1,509 tasks across VirtualHome and Habitat 2.0, categorized into ultra-short, short, medium, and long tasks. Experimental results demonstrate that SPO significantly improves reasoning quality and final decision accuracy, outperforming prior methods on long-horizon tasks and underscoring the effectiveness of preference-driven optimization in vision-language task planning. Specifically, SPO achieves a +5.98% GCR and +4.68% SR improvement in VirtualHome and a +3.30% GCR and +2.11% SR improvement in Habitat over the best-performing baselines.
comment: 18 pages
♻ ☆ Enhancing Vietnamese VQA through Curriculum Learning on Raw and Augmented Text Representations AAAI-25
Visual Question Answering (VQA) is a multimodal task requiring reasoning across textual and visual inputs, which becomes particularly challenging in low-resource languages like Vietnamese due to linguistic variability and the lack of high-quality datasets. Traditional methods often rely heavily on extensive annotated datasets, computationally expensive pipelines, and large pre-trained models, specifically in the domain of Vietnamese VQA, limiting their applicability in such scenarios. To address these limitations, we propose a training framework that combines a paraphrase-based feature augmentation module with a dynamic curriculum learning strategy. Explicitly, augmented samples are considered "easy" while raw samples are regarded as "hard". The framework then utilizes a mechanism that dynamically adjusts the ratio of easy to hard samples during training, progressively modifying the same dataset to increase its difficulty level. By enabling gradual adaptation to task complexity, this approach helps the Vietnamese VQA model generalize well, thus improving overall performance. Experimental results show consistent improvements on the OpenViVQA dataset and mixed outcomes on the ViVQA dataset, highlighting both the potential and challenges of our approach in advancing VQA for Vietnamese language.
comment: 10 pages, 3 figures, AAAI-25 Workshop on Document Understanding and Intelligence
♻ ☆ Pathfinder for Low-altitude Aircraft with Binary Neural Network
A prior global topological map (e.g., the OpenStreetMap, OSM) can boost the performance of autonomous mapping by a ground mobile robot. However, the prior map is usually incomplete due to lacking labeling in partial paths. To solve this problem, this paper proposes an OSM maker using airborne sensors carried by low-altitude aircraft, where the core of the OSM maker is a novel efficient pathfinder approach based on LiDAR and camera data, i.e., a binary dual-stream road segmentation model. Specifically, a multi-scale feature extraction based on the UNet architecture is implemented for images and point clouds. To reduce the effect caused by the sparsity of point cloud, an attention-guided gated block is designed to integrate image and point-cloud features. To optimize the model for edge deployment that significantly reduces storage footprint and computational demands, we propose a binarization streamline to each model component, including a variant of vision transformer (ViT) architecture as the encoder of the image branch, and new focal and perception losses to optimize the model training. The experimental results on two datasets demonstrate that our pathfinder method achieves SOTA accuracy with high efficiency in finding paths from the low-level airborne sensors, and we can create complete OSM prior maps based on the segmented road skeletons. Code and data are available at: \href{https://github.com/IMRL/Pathfinder}{https://github.com/IMRL/Pathfinder}.
♻ ☆ Deep unrolling for learning optimal spatially varying regularisation parameters for Total Generalised Variation
We extend a recently introduced deep unrolling framework for learning spatially varying regularisation parameters in inverse imaging problems to the case of Total Generalised Variation (TGV). The framework combines a deep convolutional neural network (CNN) inferring the two spatially varying TGV parameters with an unrolled algorithmic scheme that solves the corresponding variational problem. The two subnetworks are jointly trained end-to-end in a supervised fashion and as such the CNN learns to compute those parameters that drive the reconstructed images as close to the ground truth as possible. Numerical results in image denoising and MRI reconstruction show a significant qualitative and quantitative improvement compared to the best TGV scalar parameter case as well as to other approaches employing spatially varying parameters computed by unsupervised methods. We also observe that the inferred spatially varying parameter maps have a consistent structure near the image edges, asking for further theoretical investigations. In particular, the parameter that weighs the first-order TGV term has a triple-edge structure with alternating high-low-high values whereas the one that weighs the second-order term attains small values in a large neighbourhood around the edges.
♻ ☆ InfoDisent: Explainability of Image Classification Models by Information Disentanglement
In this work, we introduce InfoDisent, a hybrid approach to explainability based on the information bottleneck principle. InfoDisent enables the disentanglement of information in the final layer of any pretrained model into atomic concepts, which can be interpreted as prototypical parts. This approach merges the flexibility of post-hoc methods with the concept-level modeling capabilities of self-explainable neural networks, such as ProtoPNets. We demonstrate the effectiveness of InfoDisent through computational experiments and user studies across various datasets using modern backbones such as ViTs and convolutional networks. Notably, InfoDisent generalizes the prototypical parts approach to novel domains (ImageNet).
♻ ☆ A Backbone for Long-Horizon Robot Task Understanding
End-to-end robot learning, particularly for long-horizon tasks, often results in unpredictable outcomes and poor generalization. To address these challenges, we propose a novel Therblig-Based Backbone Framework (TBBF) as a fundamental structure to enhance interpretability, data efficiency, and generalization in robotic systems. TBBF utilizes expert demonstrations to enable therblig-level task decomposition, facilitate efficient action-object mapping, and generate adaptive trajectories for new scenarios. The approach consists of two stages: offline training and online testing. During the offline training stage, we developed the Meta-RGate SynerFusion (MGSF) network for accurate therblig segmentation across various tasks. In the online testing stage, after a one-shot demonstration of a new task is collected, our MGSF network extracts high-level knowledge, which is then encoded into the image using Action Registration (ActionREG). Additionally, Large Language Model (LLM)-Alignment Policy for Visual Correction (LAP-VC) is employed to ensure precise action registration, facilitating trajectory transfer in novel robot scenarios. Experimental results validate these methods, achieving 94.37% recall in therblig segmentation and success rates of 94.4% and 80% in real-world online robot testing for simple and complex scenarios, respectively. Supplementary material is available at: https://sites.google.com/view/therbligsbasedbackbone/home
comment: 8 pages, 8 figures. This work has been published by IEEE Robotics and Automation Letters (RA-L)
♻ ☆ DongbaMIE: A Multimodal Information Extraction Dataset for Evaluating Semantic Understanding of Dongba Pictograms
Dongba pictographs are the only pictographs still in use in the world. They have pictorial ideographic features, and their symbols carry rich cultural and contextual information. Due to the lack of relevant datasets, existing research has difficulty in advancing the study of semantic understanding of Dongba pictographs. To this end, we propose DongbaMIE, the first multimodal dataset for semantic understanding and extraction of Dongba pictographs. The dataset consists of Dongba pictograph images and their corresponding Chinese semantic annotations. It contains 23,530 sentence-level and 2,539 paragraph-level images, covering four semantic dimensions: objects, actions, relations, and attributes. We systematically evaluate the GPT-4o, Gemini-2.0, and Qwen2-VL models. Experimental results show that the F1 scores of GPT-4o and Gemini in the best object extraction are only 3.16 and 3.11 respectively. The F1 score of Qwen2-VL after supervised fine-tuning is only 11.49. These results suggest that current large multimodal models still face significant challenges in accurately recognizing the diverse semantic information in Dongba pictographs. The dataset can be obtained from this URL.
♻ ☆ Federated Learning With Individualized Privacy Through Client Sampling ICML
With growing concerns about user data collection, individualized privacy has emerged as a promising solution to balance protection and utility by accounting for diverse user privacy preferences. Instead of enforcing a uniform level of anonymization for all users, this approach allows individuals to choose privacy settings that align with their comfort levels. Building on this idea, we propose an adapted method for enabling Individualized Differential Privacy (IDP) in Federated Learning (FL) by handling clients according to their personal privacy preferences. By extending the SAMPLE algorithm from centralized settings to FL, we calculate client-specific sampling rates based on their heterogeneous privacy budgets and integrate them into a modified IDP-FedAvg algorithm. We test this method under realistic privacy distributions and multiple datasets. The experimental results demonstrate that our approach achieves clear improvements over uniform DP baselines, reducing the trade-off between privacy and utility. Compared to the alternative SCALE method in related work, which assigns differing noise scales to clients, our method performs notably better. However, challenges remain for complex tasks with non-i.i.d. data, primarily stemming from the constraints of the decentralized setting.
comment: Accepted at 10th International Conference on Machine Learning Technologies (ICMLT 2025)
♻ ☆ Modulating CNN Features with Pre-Trained ViT Representations for Open-Vocabulary Object Detection
Owing to large-scale image-text contrastive training, pre-trained vision language model (VLM) like CLIP shows superior open-vocabulary recognition ability. Most existing open-vocabulary object detectors attempt to utilize the pre-trained VLMs to attain generalized representation. F-ViT uses the pre-trained visual encoder as the backbone network and freezes it during training. However, its frozen backbone doesn't benefit from the labeled data to strengthen the representation for detection. Therefore, we propose a novel two-branch backbone network, named as \textbf{V}iT-Feature-\textbf{M}odulated Multi-Scale \textbf{C}onvolutional Network (VMCNet), which consists of a trainable convolutional branch, a frozen pre-trained ViT branch and a VMC module. The trainable CNN branch could be optimized with labeled data while the frozen pre-trained ViT branch could keep the representation ability derived from large-scale pre-training. Then, the proposed VMC module could modulate the multi-scale CNN features with the representations from ViT branch. With this proposed mixed structure, the detector is more likely to discover objects of novel categories. Evaluated on two popular benchmarks, our method boosts the detection performance on novel category and outperforms state-of-the-art methods. On OV-COCO, the proposed method achieves 44.3 AP$_{50}^{\mathrm{novel}}$ with ViT-B/16 and 48.5 AP$_{50}^{\mathrm{novel}}$ with ViT-L/14. On OV-LVIS, VMCNet with ViT-B/16 and ViT-L/14 reaches 27.8 and 38.4 mAP$_{r}$.
♻ ☆ $σ$-zero: Gradient-based Optimization of $\ell_0$-norm Adversarial Examples ICLR 2025
Evaluating the adversarial robustness of deep networks to gradient-based attacks is challenging. While most attacks consider $\ell_2$- and $\ell_\infty$-norm constraints to craft input perturbations, only a few investigate sparse $\ell_1$- and $\ell_0$-norm attacks. In particular, $\ell_0$-norm attacks remain the least studied due to the inherent complexity of optimizing over a non-convex and non-differentiable constraint. However, evaluating adversarial robustness under these attacks could reveal weaknesses otherwise left untested with more conventional $\ell_2$- and $\ell_\infty$-norm attacks. In this work, we propose a novel $\ell_0$-norm attack, called $\sigma$-zero, which leverages a differentiable approximation of the $\ell_0$ norm to facilitate gradient-based optimization, and an adaptive projection operator to dynamically adjust the trade-off between loss minimization and perturbation sparsity. Extensive evaluations using MNIST, CIFAR10, and ImageNet datasets, involving robust and non-robust models, show that $\sigma$\texttt{-zero} finds minimum $\ell_0$-norm adversarial examples without requiring any time-consuming hyperparameter tuning, and that it outperforms all competing sparse attacks in terms of success rate, perturbation size, and efficiency.
comment: Paper accepted at International Conference on Learning Representations (ICLR 2025). Code available at https://github.com/sigma0-advx/sigma-zero
♻ ☆ VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models
In this paper, we propose a novel framework for solving high-definition video inverse problems using latent image diffusion models. Building on recent advancements in spatio-temporal optimization for video inverse problems using image diffusion models, our approach leverages latent-space diffusion models to achieve enhanced video quality and resolution. To address the high computational demands of processing high-resolution frames, we introduce a pseudo-batch consistent sampling strategy, allowing efficient operation on a single GPU. Additionally, to improve temporal consistency, we present pseudo-batch inversion, an initialization technique that incorporates informative latents from the measurement. By integrating with SDXL, our framework achieves state-of-the-art video reconstruction across a wide range of spatio-temporal inverse problems, including complex combinations of frame averaging and various spatial degradations, such as deblurring, super-resolution, and inpainting. Unlike previous methods, our approach supports multiple aspect ratios (landscape, vertical, and square) and delivers HD-resolution reconstructions (exceeding 1280x720) in under 6 seconds per frame on a single NVIDIA 4090 GPU.
comment: Project page: https://vision-xl.github.io/
♻ ☆ No More Sliding Window: Efficient 3D Medical Image Segmentation with Differentiable Top-k Patch Sampling
3D models surpass 2D models in CT/MRI segmentation by effectively capturing inter-slice relationships. However, the added depth dimension substantially increases memory consumption. While patch-based training alleviates memory constraints, it significantly slows down the inference speed due to the sliding window (SW) approach. We propose No-More-Sliding-Window (NMSW), a novel end-to-end trainable framework that enhances the efficiency of generic 3D segmentation backbone during an inference step by eliminating the need for SW. NMSW employs a differentiable Top-k module to selectively sample only the most relevant patches, thereby minimizing redundant computations. When patch-level predictions are insufficient, the framework intelligently leverages coarse global predictions to refine results. Evaluated across 3 tasks using 3 segmentation backbones, NMSW achieves competitive accuracy compared to SW inference while significantly reducing computational complexity by 91% (88.0 to 8.00 TMACs). Moreover, it delivers a 9.1x faster inference on the H100 GPU (99.0 to 8.3 sec) and a 11.1x faster inference on the Xeon Gold CPU (2110 to 189 sec). NMSW is model-agnostic, further boosting efficiency when integrated with any existing efficient segmentation backbones.
♻ ☆ OmniGuard: Hybrid Manipulation Localization via Augmented Versatile Deep Image Watermarking CVPR 2025
With the rapid growth of generative AI and its widespread application in image editing, new risks have emerged regarding the authenticity and integrity of digital content. Existing versatile watermarking approaches suffer from trade-offs between tamper localization precision and visual quality. Constrained by the limited flexibility of previous framework, their localized watermark must remain fixed across all images. Under AIGC-editing, their copyright extraction accuracy is also unsatisfactory. To address these challenges, we propose OmniGuard, a novel augmented versatile watermarking approach that integrates proactive embedding with passive, blind extraction for robust copyright protection and tamper localization. OmniGuard employs a hybrid forensic framework that enables flexible localization watermark selection and introduces a degradation-aware tamper extraction network for precise localization under challenging conditions. Additionally, a lightweight AIGC-editing simulation layer is designed to enhance robustness across global and local editing. Extensive experiments show that OmniGuard achieves superior fidelity, robustness, and flexibility. Compared to the recent state-of-the-art approach EditGuard, our method outperforms it by 4.25dB in PSNR of the container image, 20.7% in F1-Score under noisy conditions, and 14.8% in average bit accuracy.
comment: Accepted by CVPR 2025
♻ ☆ MIAdapt: Source-free Few-shot Domain Adaptive Object Detection for Microscopic Images
Existing generic unsupervised domain adaptation approaches require access to both a large labeled source dataset and a sufficient unlabeled target dataset during adaptation. However, collecting a large dataset, even if unlabeled, is a challenging and expensive endeavor, especially in medical imaging. In addition, constraints such as privacy issues can result in cases where source data is unavailable. Taking in consideration these challenges, we propose MIAdapt, an adaptive approach for Microscopic Imagery Adaptation as a solution for Source-free Few-shot Domain Adaptive Object detection (SF-FSDA). We also define two competitive baselines (1) Faster-FreeShot and (2) MT-FreeShot. Extensive experiments on the challenging M5-Malaria and Raabin-WBC datasets validate the effectiveness of MIAdapt. Without using any image from the source domain MIAdapt surpasses state-of-the-art source-free UDA (SF-UDA) methods by +21.3% mAP and few-shot domain adaptation (FSDA) approaches by +4.7% mAP on Raabin-WBC. Our code and models will be publicly available.
comment: 6 pages, 5 figures
♻ ☆ Indirect Gradient Matching for Adversarial Robust Distillation ICLR 2025
Adversarial training significantly improves adversarial robustness, but superior performance is primarily attained with large models. This substantial performance gap for smaller models has spurred active research into adversarial distillation (AD) to mitigate the difference. Existing AD methods leverage the teacher's logits as a guide. In contrast to these approaches, we aim to transfer another piece of knowledge from the teacher, the input gradient. In this paper, we propose a distillation module termed Indirect Gradient Distillation Module (IGDM) that indirectly matches the student's input gradient with that of the teacher. Experimental results show that IGDM seamlessly integrates with existing AD methods, significantly enhancing their performance. Particularly, utilizing IGDM on the CIFAR-100 dataset improves the AutoAttack accuracy from 28.06% to 30.32% with the ResNet-18 architecture and from 26.18% to 29.32% with the MobileNetV2 architecture when integrated into the SOTA method without additional data augmentation.
comment: ICLR 2025
♻ ☆ Drag Your Gaussian: Effective Drag-Based Editing with Score Distillation for 3D Gaussian Splatting
Recent advancements in 3D scene editing have been propelled by the rapid development of generative models. Existing methods typically utilize generative models to perform text-guided editing on 3D representations, such as 3D Gaussian Splatting (3DGS). However, these methods are often limited to texture modifications and fail when addressing geometric changes, such as editing a character's head to turn around. Moreover, such methods lack accurate control over the spatial position of editing results, as language struggles to precisely describe the extent of edits. To overcome these limitations, we introduce DYG, an effective 3D drag-based editing method for 3D Gaussian Splatting. It enables users to conveniently specify the desired editing region and the desired dragging direction through the input of 3D masks and pairs of control points, thereby enabling precise control over the extent of editing. DYG integrates the strengths of the implicit triplane representation to establish the geometric scaffold of the editing results, effectively overcoming suboptimal editing outcomes caused by the sparsity of 3DGS in the desired editing regions. Additionally, we incorporate a drag-based Latent Diffusion Model into our method through the proposed Drag-SDS loss function, enabling flexible, multi-view consistent, and fine-grained editing. Extensive experiments demonstrate that DYG conducts effective drag-based editing guided by control point prompts, surpassing other baselines in terms of editing effect and quality, both qualitatively and quantitatively. Visit our project page at https://quyans.github.io/Drag-Your-Gaussian.
comment: Visit our project page at https://quyans.github.io/Drag-Your-Gaussian
♻ ☆ GIFT: Unlocking Full Potential of Labels in Distilled Dataset at Near-zero Cost
Recent advancements in dataset distillation have demonstrated the significant benefits of employing soft labels generated by pre-trained teacher models. In this paper, we introduce a novel perspective by emphasizing the full utilization of labels. We first conduct a comprehensive comparison of various loss functions for soft label utilization in dataset distillation, revealing that the model trained on the synthetic dataset exhibits high sensitivity to the choice of loss function for soft label utilization. This finding highlights the necessity of a universal loss function for training models on synthetic datasets. Building on these insights, we introduce an extremely simple yet surprisingly effective plug-and-play approach, GIFT, which encompasses soft label refinement and a cosine similarity-based loss function to efficiently leverage full label information. Extensive experiments indicate that GIFT consistently enhances state-of-the-art dataset distillation methods across various dataset scales, without incurring additional computational costs. Importantly, GIFT significantly enhances cross-optimizer generalization, an area previously overlooked. For instance, on ImageNet-1K with IPC = 10, GIFT enhances the state-of-the-art method RDED by 30.8% in cross-optimizer generalization. Our code is available at https://github.com/LINs-lab/GIFT.
comment: https://github.com/LINs-lab/GIFT
♻ ☆ Manta: Enhancing Mamba for Few-Shot Action Recognition of Long Sub-Sequence AAAI 2025
In few-shot action recognition (FSAR), long sub-sequences of video naturally express entire actions more effectively. However, the high computational complexity of mainstream Transformer-based methods limits their application. Recent Mamba demonstrates efficiency in modeling long sequences, but directly applying Mamba to FSAR overlooks the importance of local feature modeling and alignment. Moreover, long sub-sequences within the same class accumulate intra-class variance, which adversely impacts FSAR performance. To solve these challenges, we propose a Matryoshka MAmba and CoNtrasTive LeArning framework (Manta). Firstly, the Matryoshka Mamba introduces multiple Inner Modules to enhance local feature representation, rather than directly modeling global features. An Outer Module captures dependencies of timeline between these local features for implicit temporal alignment. Secondly, a hybrid contrastive learning paradigm, combining both supervised and unsupervised methods, is designed to mitigate the negative effects of intra-class variance accumulation. The Matryoshka Mamba and the hybrid contrastive learning paradigm operate in two parallel branches within Manta, enhancing Mamba for FSAR of long sub-sequence. Manta achieves new state-of-the-art performance on prominent benchmarks, including SSv2, Kinetics, UCF101, and HMDB51. Extensive empirical studies prove that Manta significantly improves FSAR of long sub-sequence from multiple perspectives.
comment: Accepted by AAAI 2025
♻ ☆ Locate Anything on Earth: Advancing Open-Vocabulary Object Detection for Remote Sensing Community
Object detection, particularly open-vocabulary object detection, plays a crucial role in Earth sciences, such as environmental monitoring, natural disaster assessment, and land-use planning. However, existing open-vocabulary detectors, primarily trained on natural-world images, struggle to generalize to remote sensing images due to a significant data domain gap. Thus, this paper aims to advance the development of open-vocabulary object detection in remote sensing community. To achieve this, we first reformulate the task as Locate Anything on Earth (LAE) with the goal of detecting any novel concepts on Earth. We then developed the LAE-Label Engine which collects, auto-annotates, and unifies up to 10 remote sensing datasets creating the LAE-1M - the first large-scale remote sensing object detection dataset with broad category coverage. Using the LAE-1M, we further propose and train the novel LAE-DINO Model, the first open-vocabulary foundation object detector for the LAE task, featuring Dynamic Vocabulary Construction (DVC) and Visual-Guided Text Prompt Learning (VisGT) modules. DVC dynamically constructs vocabulary for each training batch, while VisGT maps visual features to semantic space, enhancing text features. We comprehensively conduct experiments on established remote sensing benchmark DIOR, DOTAv2.0, as well as our newly introduced 80-class LAE-80C benchmark. Results demonstrate the advantages of the LAE-1M dataset and the effectiveness of the LAE-DINO method.
comment: 15 pages, 11 figures
♻ ☆ SSL4EO-S12 v1.1: A Multimodal, Multiseasonal Dataset for Pretraining, Updated
This technical report presents SSL4EO-S12 v1.1, a multimodal, multitemporal Earth Observation dataset designed for pretraining large-scale foundation models. Building on the success of SSL4EO-S12 v1.0, the new version addresses the previous challenges of data misalignment and a limited data structure for low-barrier, analysis-ready EO processing. SSL4EO-S12 v1.1 covers the world's 10,000 largest cities and its surroundings within a 50 km radius across four seasons, resulting in a diverse collection of nearly one million patches. SSL4EO-S12 v1.1 packages the data in Zarr file format for cloud-efficient loading and representation of meta-information such as including cloud masks and geolocation. Released under the CC-BY-4.0 license, SSL4EO-S12 v1.1 facilitates open research and provides a robust foundation for future advancements in self-supervised learning and geospatial analysis. The dataset is available online through https://datapub.fz-juelich.de/ssl4eo-s12, and we provided additional resources at https://github.com/DLR-MF-DAS/SSL4EO-S12-v1.1.
♻ ☆ StoryTeller: Improving Long Video Description through Global Audio-Visual Character Identification
Existing large vision-language models (LVLMs) are largely limited to processing short, seconds-long videos and struggle with generating coherent descriptions for extended video spanning minutes or more. Long video description introduces new challenges, such as consistent character identification and plot-level descriptions incorporating both visual and audio information. To address these, we figure out audio-visual character identification, matching character names to each dialogue, as a key factor. We propose StoryTeller, a system for generating dense descriptions of long videos, incorporating both low-level visual concepts and high-level plot information. StoryTeller uses a multimodal large language model that integrates visual, audio, and text modalities to perform audio-visual character identification on minute-long video clips. The results are then fed into a LVLM to enhance consistency of video description. We validate our approach on movie description tasks and introduce MovieStory101, a dataset with dense descriptions for three-minute movie clips. To evaluate long video descriptions, we create StoryQA, a large set of multiple-choice questions for MovieStory101 test set. We assess descriptions by inputting them into GPT-4 to answer these questions, using accuracy as an automatic evaluation metric. Experiments show that StoryTeller outperforms all open and closed-source baselines on StoryQA, achieving 9.5% higher accuracy than the strongest baseline, Gemini-1.5-pro, and demonstrating a +15.56% advantage in human side-by-side evaluations. Additionally, incorporating audio-visual character identification from StoryTeller improves the performance of all video description models, with Gemini-1.5-pro and GPT-4o showing relative improvement of 5.5% and 13.0%, respectively, in accuracy on StoryQA.
♻ ☆ Rethinking Weight-Averaged Model-merging
Model-merging has emerged as a powerful approach in deep learning, capable of enhancing model performance without any training. However, the underlying mechanisms that explain its effectiveness remain largely unexplored. In this paper, we investigate this technique from three novel perspectives to empirically provide deeper insights into why and how weight-averaged model-merging works: (1) we examine the intrinsic patterns captured by the learning of the model weights, through the visualizations of their patterns on several datasets, showing that these weights often encode structured and interpretable patterns and that is the essential why model-merging can work; (2) we mathematically and empirically investigate model ensemble merging strategies based on averaging on weights versus averaging on features, providing detailed analyses across diverse architectures and datasets; and (3) we explore the impact on model-merging prediction stability in terms of changing the parameter magnitude, revealing insights into the way of weight averaging works as regularization by showing the robustness across different parameter scales. Our findings shed light on the "black box" of weight-averaged model-merging, offering valuable insights and practical recommendations that advance the model-merging process. The code is available at https://github.com/billhhh/Rethink-Merge.
♻ ☆ Meta-Learned Modality-Weighted Knowledge Distillation for Robust Multi-Modal Learning with Missing Data
In multi-modal learning, some modalities are more influential than others, and their absence can have a significant impact on classification/segmentation accuracy. Addressing this challenge, we propose a novel approach called Meta-learned Modality-weighted Knowledge Distillation (MetaKD), which enables multi-modal models to maintain high accuracy even when key modalities are missing. MetaKD adaptively estimates the importance weight of each modality through a meta-learning process. These learned importance weights guide a pairwise modality-weighted knowledge distillation process, allowing high-importance modalities to transfer knowledge to lower-importance ones, resulting in robust performance despite missing inputs. Unlike previous methods in the field, which are often task-specific and require significant modifications, our approach is designed to work in multiple tasks (e.g., segmentation and classification) with minimal adaptation. Experimental results on five prevalent datasets, including three Brain Tumor Segmentation datasets (BraTS2018, BraTS2019 and BraTS2020), the Alzheimer's Disease Neuroimaging Initiative (ADNI) classification dataset and the Audiovision-MNIST classification dataset, demonstrate the proposed model is able to outperform the compared models by a large margin. The code is available at https://github.com/billhhh/MetaKD.
♻ ☆ Efficient Masked AutoEncoder for Video Object Counting and A Large-Scale Benchmark ICLR25
The dynamic imbalance of the fore-background is a major challenge in video object counting, which is usually caused by the sparsity of target objects. This remains understudied in existing works and often leads to severe under-/over-prediction errors. To tackle this issue in video object counting, we propose a density-embedded Efficient Masked Autoencoder Counting (E-MAC) framework in this paper. To empower the model's representation ability on density regression, we develop a new $\mathtt{D}$ensity-$\mathtt{E}$mbedded $\mathtt{M}$asked m$\mathtt{O}$deling ($\mathtt{DEMO}$) method, which first takes the density map as an auxiliary modality to perform multimodal self-representation learning for image and density map. Although $\mathtt{DEMO}$ contributes to effective cross-modal regression guidance, it also brings in redundant background information, making it difficult to focus on the foreground regions. To handle this dilemma, we propose an efficient spatial adaptive masking derived from density maps to boost efficiency. Meanwhile, we employ an optical flow-based temporal collaborative fusion strategy to effectively capture the dynamic variations across frames, aligning features to derive multi-frame density residuals. The counting accuracy of the current frame is boosted by harnessing the information from adjacent frames. In addition, considering that most existing datasets are limited to human-centric scenarios, we first propose a large video bird counting dataset, DroneBird, in natural scenarios for migratory bird protection. Extensive experiments on three crowd datasets and our \textit{DroneBird} validate our superiority against the counterparts. The code and dataset are available.
comment: ICLR25
♻ ☆ LaVin-DiT: Large Vision Diffusion Transformer CVPR 2025
This paper presents the Large Vision Diffusion Transformer (LaVin-DiT), a scalable and unified foundation model designed to tackle over 20 computer vision tasks in a generative framework. Unlike existing large vision models directly adapted from natural language processing architectures, which rely on less efficient autoregressive techniques and disrupt spatial relationships essential for vision data, LaVin-DiT introduces key innovations to optimize generative performance for vision tasks. First, to address the high dimensionality of visual data, we incorporate a spatial-temporal variational autoencoder that encodes data into a continuous latent space. Second, for generative modeling, we develop a joint diffusion transformer that progressively produces vision outputs. Third, for unified multi-task training, in-context learning is implemented. Input-target pairs serve as task context, which guides the diffusion transformer to align outputs with specific tasks within the latent space. During inference, a task-specific context set and test data as queries allow LaVin-DiT to generalize across tasks without fine-tuning. Trained on extensive vision datasets, the model is scaled from 0.1B to 3.4B parameters, demonstrating substantial scalability and state-of-the-art performance across diverse vision tasks. This work introduces a novel pathway for large vision foundation models, underscoring the promising potential of diffusion transformers. The code and models are available.
comment: 37 pages, 30 figures, 4 tables. Accepted by CVPR 2025
♻ ☆ Self-Adaptive Gamma Context-Aware SSM-based Model for Metal Defect Detection
Metal defect detection is critical in industrial quality assurance, yet existing methods struggle with grayscale variations and complex defect states, limiting its robustness. To address these challenges, this paper proposes a Self-Adaptive Gamma Context-Aware SSM-based model(GCM-DET). This advanced detection framework integrating a Dynamic Gamma Correction (GC) module to enhance grayscale representation and optimize feature extraction for precise defect reconstruction. A State-Space Search Management (SSM) architecture captures robust multi-scale features, effectively handling defects of varying shapes and scales. Focal Loss is employed to mitigate class imbalance and refine detection accuracy. Additionally, the CD5-DET dataset is introduced, specifically designed for port container maintenance, featuring significant grayscale variations and intricate defect patterns. Experimental results demonstrate that the proposed model achieves substantial improvements, with mAP@0.5 gains of 27.6\%, 6.6\%, and 2.6\% on the CD5-DET, NEU-DET, and GC10-DET datasets.
comment: 19 pages, 9 figures, under review
♻ ☆ DRACO-DehazeNet: An Efficient Image Dehazing Network Combining Detail Recovery and a Novel Contrastive Learning Paradigm
Image dehazing is crucial for clarifying images obscured by haze or fog, but current learning-based approaches is dependent on large volumes of training data and hence consumed significant computational power. Additionally, their performance is often inadequate under non-uniform or heavy haze. To address these challenges, we developed the Detail Recovery And Contrastive DehazeNet, which facilitates efficient and effective dehazing via a dense dilated inverted residual block and an attention-based detail recovery network that tailors enhancements to specific dehazed scene contexts. A major innovation is its ability to train effectively with limited data, achieved through a novel quadruplet loss-based contrastive dehazing paradigm. This approach distinctly separates hazy and clear image features while also distinguish lower-quality and higher-quality dehazed images obtained from each sub-modules of our network, thereby refining the dehazing process to a larger extent. Extensive tests on a variety of benchmarked haze datasets demonstrated the superiority of our approach. The code repository for this work is available at https://github.com/GreedYLearner1146/DRACO-DehazeNet.
comment: Once the paper is accepted and published, the copyright will be transferred to the corresponding journal
♻ ☆ Novel Pipeline for Diagnosing Acute Lymphoblastic Leukemia Sensitive to Related Biomarkers
Acute Lymphoblastic Leukemia (ALL) is one of the most common types of childhood blood cancer. The quick start of the treatment process is critical to saving the patient's life, and for this reason, early diagnosis of this disease is essential. Examining the blood smear images of these patients is one of the methods used by expert doctors to diagnose this disease. Deep learning-based methods have numerous applications in medical fields, as they have significantly advanced in recent years. ALL diagnosis is not an exception in this field, and several machine learning-based methods for this problem have been proposed. In previous methods, high diagnostic accuracy was reported, but our work showed that this alone is not sufficient, as it can lead to models taking shortcuts and not making meaningful decisions. This issue arises due to the small size of medical training datasets. To address this, we constrained our model to follow a pipeline inspired by experts' work. We also demonstrated that, since a judgement based on only one image is insufficient, redefining the problem as a multiple-instance learning problem is necessary for achieving a practical result. Our model is the first to provide a solution to this problem in a multiple-instance learning setup. We introduced a novel pipeline for diagnosing ALL that approximates the process used by hematologists, is sensitive to disease biomarkers, and achieves an accuracy of 96.15%, an F1-score of 94.24%, a sensitivity of 97.56%, and a specificity of 90.91% on ALL IDB 1. Our method was further evaluated on an out-of-distribution dataset, which posed a challenging test and had acceptable performance. Notably, our model was trained on a relatively small dataset, highlighting the potential for our approach to be applied to other medical datasets with limited data availability.
♻ ☆ Dur360BEV: A Real-world 360-degree Single Camera Dataset and Benchmark for Bird-Eye View Mapping in Autonomous Driving
We present Dur360BEV, a novel spherical camera autonomous driving dataset equipped with a high-resolution 128-channel 3D LiDAR and a RTK-refined GNSS/INS system, along with a benchmark architecture designed to generate Bird-Eye-View (BEV) maps using only a single spherical camera. This dataset and benchmark address the challenges of BEV generation in autonomous driving, particularly by reducing hardware complexity through the use of a single 360-degree camera instead of multiple perspective cameras. Within our benchmark architecture, we propose a novel spherical-image-to-BEV module that leverages spherical imagery and a refined sampling strategy to project features from 2D to 3D. Our approach also includes an innovative application of focal loss, specifically adapted to address the extreme class imbalance often encountered in BEV segmentation tasks, that demonstrates improved segmentation performance on the Dur360BEV dataset. The results show that our benchmark not only simplifies the sensor setup but also achieves competitive performance.
♻ ☆ DexMimicGen: Automated Data Generation for Bimanual Dexterous Manipulation via Imitation Learning ICRA 2025
Imitation learning from human demonstrations is an effective means to teach robots manipulation skills. But data acquisition is a major bottleneck in applying this paradigm more broadly, due to the amount of cost and human effort involved. There has been significant interest in imitation learning for bimanual dexterous robots, like humanoids. Unfortunately, data collection is even more challenging here due to the challenges of simultaneously controlling multiple arms and multi-fingered hands. Automated data generation in simulation is a compelling, scalable alternative to fuel this need for data. To this end, we introduce DexMimicGen, a large-scale automated data generation system that synthesizes trajectories from a handful of human demonstrations for humanoid robots with dexterous hands. We present a collection of simulation environments in the setting of bimanual dexterous manipulation, spanning a range of manipulation behaviors and different requirements for coordination among the two arms. We generate 21K demos across these tasks from just 60 source human demos and study the effect of several data generation and policy learning decisions on agent performance. Finally, we present a real-to-sim-to-real pipeline and deploy it on a real-world humanoid can sorting task. Generated datasets, simulation environments and additional results are at https://dexmimicgen.github.io/
comment: ICRA 2025. Project website: https://dexmimicgen.github.io/
♻ ☆ LangGas: Introducing Language in Selective Zero-Shot Background Subtraction for Semi-Transparent Gas Leak Detection with a New Dataset
Gas leakage poses a significant hazard that requires prevention. Traditionally, human inspection has been used for detection, a slow and labour-intensive process. Recent research has applied machine learning techniques to this problem, yet there remains a shortage of high-quality, publicly available datasets. This paper introduces a synthetic dataset featuring diverse backgrounds, interfering foreground objects, diverse leak locations, and precise segmentation ground truth. We propose a zero-shot method that combines background subtraction, zero-shot object detection, filtering, and segmentation to leverage this dataset. Experimental results indicate that our approach significantly outperforms baseline methods based solely on background subtraction and zero-shot object detection with segmentation, reaching an IoU of 69\% overall. We also present an analysis of various prompt configurations and threshold settings to provide deeper insights into the performance of our method. The code and dataset will be released after publication.
♻ ☆ Comparing Deep Neural Network for Multi-Label ECG Diagnosis From Scanned ECG
Automated ECG diagnosis has seen significant advancements with deep learning techniques, but real-world applications still face challenges when dealing with scanned paper ECGs. In this study, we explore multi-label classification of ECGs extracted from scanned images, moving beyond traditional binary classification (normal/abnormal). We evaluate the performance of multiple deep neural network architectures, including AlexNet, VGG, ResNet, and Vision Transformer, on scanned ECG datasets. Our comparative analysis examines model accuracy, robustness to image artifacts, and generalizability across different ECG conditions. Additionally, we investigate whether ECG signals extracted from scanned images retain sufficient diagnostic information for reliable automated classification. The findings highlight the strengths and limitations of each architecture, providing insights into the feasibility of image-based ECG diagnosis and its potential integration into clinical workflows.
♻ ☆ Human Motion Instruction Tuning CVPR 2025
This paper presents LLaMo (Large Language and Human Motion Assistant), a multimodal framework for human motion instruction tuning. In contrast to conventional instruction-tuning approaches that convert non-linguistic inputs, such as video or motion sequences, into language tokens, LLaMo retains motion in its native form for instruction tuning. This method preserves motion-specific details that are often diminished in tokenization, thereby improving the model's ability to interpret complex human behaviors. By processing both video and motion data alongside textual inputs, LLaMo enables a flexible, human-centric analysis. Experimental evaluations across high-complexity domains, including human behaviors and professional activities, indicate that LLaMo effectively captures domain-specific knowledge, enhancing comprehension and prediction in motion-intensive scenarios. We hope LLaMo offers a foundation for future multimodal AI systems with broad applications, from sports analytics to behavioral prediction. Our code and models are available on the project website: https://github.com/ILGLJ/LLaMo.
comment: Accepted by CVPR 2025
♻ ☆ DreamText: High Fidelity Scene Text Synthesis
Scene text synthesis involves rendering specified texts onto arbitrary images. Current methods typically formulate this task in an end-to-end manner but lack effective character-level guidance during training. Besides, their text encoders, pre-trained on a single font type, struggle to adapt to the diverse font styles encountered in practical applications. Consequently, these methods suffer from character distortion, repetition, and absence, particularly in polystylistic scenarios. To this end, this paper proposes DreamText for high-fidelity scene text synthesis. Our key idea is to reconstruct the diffusion training process, introducing more refined guidance tailored to this task, to expose and rectify the model's attention at the character level and strengthen its learning of text regions. This transformation poses a hybrid optimization challenge, involving both discrete and continuous variables. To effectively tackle this challenge, we employ a heuristic alternate optimization strategy. Meanwhile, we jointly train the text encoder and generator to comprehensively learn and utilize the diverse font present in the training dataset. This joint training is seamlessly integrated into the alternate optimization process, fostering a synergistic relationship between learning character embedding and re-estimating character attention. Specifically, in each step, we first encode potential character-generated position information from cross-attention maps into latent character masks. These masks are then utilized to update the representation of specific characters in the current step, which, in turn, enables the generator to correct the character's attention in the subsequent steps. Both qualitative and quantitative results demonstrate the superiority of our method to the state of the art.
comment: Code: https://github.com/CodeGoat24/DreamText, Project page: https://codegoat24.github.io/DreamText/
♻ ☆ TractCloud-FOV: Deep Learning-based Robust Tractography Parcellation in Diffusion MRI with Incomplete Field of View
Tractography parcellation classifies streamlines reconstructed from diffusion MRI into anatomically defined fiber tracts for clinical and research applications. However, clinical scans often have incomplete fields of view (FOV) where brain regions are partially imaged, leading to partial or truncated fiber tracts. To address this challenge, we introduce TractCloud-FOV, a deep learning framework that robustly parcellates tractography under conditions of incomplete FOV. We propose a novel training strategy, FOV-Cut Augmentation (FOV-CA), in which we synthetically cut tractograms to simulate a spectrum of real-world inferior FOV cutoff scenarios. This data augmentation approach enriches the training set with realistic truncated streamlines, enabling the model to achieve superior generalization. We evaluate the proposed TractCloud-FOV on both synthetically cut tractography and two real-life datasets with incomplete FOV. TractCloud-FOV significantly outperforms several state-of-the-art methods on all testing datasets in terms of streamline classification accuracy, generalization ability, tract anatomical depiction, and computational efficiency. Overall, TractCloud-FOV achieves efficient and consistent tractography parcellation in diffusion MRI with incomplete FOV.
♻ ☆ Visual Description Grounding Reduces Hallucinations and Boosts Reasoning in LVLMs ICLR 2025
Large Vision-Language Models (LVLMs) often produce responses that misalign with factual information, a phenomenon known as hallucinations. While hallucinations are well-studied, the exact causes behind them remain underexplored. In this paper, we first investigate the root causes of hallucinations in LVLMs. Our findings reveal that existing mitigation techniques primarily reduce hallucinations for visual recognition prompts-those that require simple descriptions of visual elements-but fail for cognitive prompts that demand deliberate reasoning. We identify the core issue as a lack of true visual perception in LVLMs: although they can accurately recognize visual elements, they struggle to fully interpret these elements in the context of the input prompt and effectively link this recognition to their internal knowledge, which is critical for reasoning. To address this gap, we introduce Visual Description Grounded Decoding (VDGD), a simple, robust, and training-free method designed to enhance visual perception and improve reasoning capabilities in LVLMs. VDGD works by first generating a detailed description of the image and appending it as a prefix to the instruction. During response generation, tokens are sampled based on their KL divergence to the description, favoring candidates with lower divergence. Experimental results on multiple visual reasoning benchmarks and LVLMs demonstrate that VDGD consistently outperforms existing baselines 2% - 33%. Finally, we introduce VaLLu, a benchmark designed for comprehensive evaluation of the cognitive capabilities of LVLMs.
comment: Accepted to ICLR 2025. Project: https://sreyan88.github.io/VDGD/
♻ ☆ Iterative Flow Matching -- Path Correction and Gradual Refinement for Enhanced Generative Modeling
Generative models for image generation are now commonly used for a wide variety of applications, ranging from guided image generation for entertainment to solving inverse problems. Nonetheless, training a generator is a non-trivial feat that requires fine-tuning and can lead to so-called hallucinations, that is, the generation of images that are unrealistic. In this work, we explore image generation using flow matching. We explain and demonstrate why flow matching can generate hallucinations, and propose an iterative process to improve the generation process. Our iterative process can be integrated into virtually $\textit{any}$ generative modeling technique, thereby enhancing the performance and robustness of image synthesis systems.
comment: 17 pages, 8 figures
♻ ☆ LesionDiffusion: Towards Text-controlled General Lesion Synthesis
Fully-supervised lesion recognition methods in medical imaging face challenges due to the reliance on large annotated datasets, which are expensive and difficult to collect. To address this, synthetic lesion generation has become a promising approach. However, existing models struggle with scalability, fine-grained control over lesion attributes, and the generation of complex structures. We propose LesionDiffusion, a text-controllable lesion synthesis framework for 3D CT imaging that generates both lesions and corresponding masks. By utilizing a structured lesion report template, our model provides greater control over lesion attributes and supports a wider variety of lesion types. We introduce a dataset of 1,505 annotated CT scans with paired lesion masks and structured reports, covering 14 lesion types across 8 organs. LesionDiffusion consists of two components: a lesion mask synthesis network (LMNet) and a lesion inpainting network (LINet), both guided by lesion attributes and image features. Extensive experiments demonstrate that LesionDiffusion significantly improves segmentation performance, with strong generalization to unseen lesion types and organs, outperforming current state-of-the-art models. Code will be available at https://github.com/HengruiTianSJTU/LesionDiffusion.
comment: 10 pages, 4 figures
♻ ☆ Shazam: Unifying Multiple Foundation Models for Advanced Computational Pathology
Foundation Models (FMs) in computational pathology (CPath) have significantly advanced the extraction of meaningful features from histopathology image datasets, achieving strong performance across various clinical tasks. Despite their impressive performance, these models often exhibit variability when applied to different tasks, prompting the need for a unified framework capable of consistently excelling across various applications. In this work, we propose Shazam, a novel framework designed to efficiently combine multiple CPath models. Unlike previous approaches that train a fixed-parameter FM, Shazam dynamically extracts and refines information from diverse FMs for each specific task. To ensure that each FM contributes effectively without dominance, a novel distillation strategy is applied, guiding the student model with features from all teacher models, which enhances its generalization ability. Experimental results on two pathology patch classification datasets demonstrate that Shazam outperforms existing CPath models and other fusion methods. Its lightweight, flexible design makes it a promising solution for improving CPath analysis in real-world settings. Code will be available at https://github.com/Tuner12/Shazam.
comment: 9 pages, 2 figures
♻ ☆ AutoBench-V: Can Large Vision-Language Models Benchmark Themselves?
Large Vision-Language Models (LVLMs) have become essential for advancing the integration of visual and linguistic information. However, the evaluation of LVLMs presents significant challenges as the evaluation benchmark always demands lots of human cost for its construction, and remains static, lacking flexibility once constructed. Even though automatic evaluation has been explored in textual modality, the visual modality remains under-explored. As a result, in this work, we address a question: "Can LVLMs themselves be used to benchmark each other in the visual automatically domain?". We introduce AutoBench-V, an automated framework for serving evaluation on demand, i.e., benchmarking LVLMs based on specific aspects of model capability. AutoBench-V leverages text-to-image models to generate relevant image samples and then utilizes LVLMs to orchestrate visual question-answering (VQA) tasks, completing the evaluation process efficiently and flexibly. Through an extensive evaluation of nine popular LVLMs across five demanded user inputs (i.e., evaluation capabilities), the framework shows effectiveness and reliability.
♻ ☆ Quantum Down Sampling Filter for Variational Auto-encoder
Variational autoencoders (VAEs) are fundamental for generative modeling and image reconstruction, yet their performance often struggles to maintain high fidelity in reconstructions. This study introduces a hybrid model, quantum variational autoencoder (Q-VAE), which integrates quantum encoding within the encoder while utilizing fully connected layers to extract meaningful representations. The decoder uses transposed convolution layers for up-sampling. The Q-VAE is evaluated against the classical VAE and the classical direct-passing VAE, which utilizes windowed pooling filters. Results on the MNIST and USPS datasets demonstrate that Q-VAE consistently outperforms classical approaches, achieving lower Fr\'echet inception distance scores, thereby indicating superior image fidelity and enhanced reconstruction quality. These findings highlight the potential of Q-VAE for high-quality synthetic data generation and improved image reconstruction in generative models.
comment: 18 pages, 13 figures
♻ ☆ Learning to Chain Operations by Routing Information Through a Global Workspace
We present a model inspired by the Global Workspace Theory that integrates specialized modules to perform a sequential reasoning task. A controller selectively routes information between modules through the workspace using a gating mechanism. This approach allows the model to chain operations by iteratively broadcasting information between specialized domains, mimicking System-2 reasoning. We evaluate the model's performance on a simple addition task, where two addends must be summed. The task can be solved by routing information sequentially through an Input module, an Increment module (multiple times), and finally an Output module. We consider two implementations of this system with increasing complexity. First, using hand-designed modules operating on one-hot digit representations, the controller (a LSTM recurrent network) learns to select the appropriate modules (input, increment, output) in the appropriate sequence. Second, we replace the hand-designed modules with learned representation modules for MNIST images and an increment module trained on the task objectives; here again, the controller learns the appropriate sequential module selection to solve the task. Finally, we show that the Global Workspace model, while having fewer parameters, outperforms LSTMs and Transformers when tested on unseen addition operations (both interpolations and extrapolations of addition operations seen during training). Our results highlight the potential of architectures inspired by the Global Workspace Theory to enhance deep learning's reasoning capabilities.
comment: 12 pages, 14 figures, submitted to a conference
♻ ☆ MBSS-T1: Model-Based Subject-Specific Self-Supervised Motion Correction for Robust Cardiac T1 Mapping
Cardiac T1 mapping is a valuable quantitative MRI technique for diagnosing diffuse myocardial diseases. Traditional methods, relying on breath-hold sequences and cardiac triggering based on an ECG signal, face challenges with patient compliance, limiting their effectiveness. Image registration can enable motion-robust cardiac T1 mapping, but inherent intensity differences between time points pose a challenge. We present MBSS-T1, a subject-specific self-supervised model for motion correction in cardiac T1 mapping. Physical constraints, implemented through a loss function comparing synthesized and motion-corrected images, enforce signal decay behavior, while anatomical constraints, applied via a Dice loss, ensure realistic deformations. The unique combination of these constraints results in motion-robust cardiac T1 mapping along the longitudinal relaxation axis. In a 5-fold experiment on a public dataset of 210 patients (STONE sequence) and an internal dataset of 19 patients (MOLLI sequence), MBSS-T1 outperformed baseline deep-learning registration methods. It achieved superior model fitting quality ($R^2$: 0.975 vs. 0.941, 0.946 for STONE; 0.987 vs. 0.982, 0.965 for MOLLI free-breathing; 0.994 vs. 0.993, 0.991 for MOLLI breath-hold), anatomical alignment (Dice: 0.89 vs. 0.84, 0.88 for STONE; 0.963 vs. 0.919, 0.851 for MOLLI free-breathing; 0.954 vs. 0.924, 0.871 for MOLLI breath-hold), and visual quality (4.33 vs. 3.38, 3.66 for STONE; 4.1 vs. 3.5, 3.28 for MOLLI free-breathing; 3.79 vs. 3.15, 2.84 for MOLLI breath-hold). MBSS-T1 enables motion-robust T1 mapping for broader patient populations, overcoming challenges such as suboptimal compliance, and facilitates free-breathing cardiac T1 mapping without requiring large annotated datasets. Our code is available at https://github.com/TechnionComputationalMRILab/MBSS-T1.
comment: Accepted and published in Medical Image Analysis
Machine Learning 152
☆ L$^2$M: Mutual Information Scaling Law for Long-Context Language Modeling
We rigorously establish a bipartite mutual information scaling law in natural language that governs long-range dependencies. This scaling law, which we show is distinct from and scales independently of the conventional two-point mutual information, is the key to understanding long-context language modeling. Using this scaling law, we formulate the Long-context Language Modeling (L$^2$M) condition, which relates a model's capacity for effective long context length modeling to the scaling of its latent state size for storing past information. Our results are validated through experiments on both transformers and state space models. This work establishes a theoretical foundation that guides the development of large language models toward longer context lengths.
comment: 29 pages, 12 figures, 1 table
☆ Enough Coin Flips Can Make LLMs Act Bayesian
Large language models (LLMs) exhibit the ability to generalize given few-shot examples in their input prompt, an emergent capability known as in-context learning (ICL). We investigate whether LLMs utilize ICL to perform structured reasoning in ways that are consistent with a Bayesian framework or rely on pattern matching. Using a controlled setting of biased coin flips, we find that: (1) LLMs often possess biased priors, causing initial divergence in zero-shot settings, (2) in-context evidence outweighs explicit bias instructions, (3) LLMs broadly follow Bayesian posterior updates, with deviations primarily due to miscalibrated priors rather than flawed updates, and (4) attention magnitude has negligible effect on Bayesian inference. With sufficient demonstrations of biased coin flips via ICL, LLMs update their priors in a Bayesian manner.
☆ Floxels: Fast Unsupervised Voxel Based Scene Flow Estimation CVPR 2025
Scene flow estimation is a foundational task for many robotic applications, including robust dynamic object detection, automatic labeling, and sensor synchronization. Two types of approaches to the problem have evolved: 1) Supervised and 2) optimization-based methods. Supervised methods are fast during inference and achieve high-quality results, however, they are limited by the need for large amounts of labeled training data and are susceptible to domain gaps. In contrast, unsupervised test-time optimization methods do not face the problem of domain gaps but usually suffer from substantial runtime, exhibit artifacts, or fail to converge to the right solution. In this work, we mitigate several limitations of existing optimization-based methods. To this end, we 1) introduce a simple voxel grid-based model that improves over the standard MLP-based formulation in multiple dimensions and 2) introduce a new multiframe loss formulation. 3) We combine both contributions in our new method, termed Floxels. On the Argoverse 2 benchmark, Floxels is surpassed only by EulerFlow among unsupervised methods while achieving comparable performance at a fraction of the computational cost. Floxels achieves a massive speedup of more than ~60 - 140x over EulerFlow, reducing the runtime from a day to 10 minutes per sequence. Over the faster but low-quality baseline, NSFP, Floxels achieves a speedup of ~14x.
comment: Accepted at CVPR 2025
☆ Predictable Scale: Part I -- Optimal Hyperparameter Scaling Law in Large Language Model Pretraining
The impressive capabilities of Large Language Models (LLMs) across diverse tasks are now well-established, yet their effective deployment necessitates careful hyperparameter optimization. Through extensive empirical studies involving grid searches across diverse configurations, we discover universal scaling laws governing these hyperparameters: optimal learning rate follows a power-law relationship with both model parameters and data sizes, while optimal batch size scales primarily with data sizes. Our analysis reveals a convex optimization landscape for hyperparameters under fixed models and data size conditions. This convexity implies an optimal hyperparameter plateau. We contribute a universal, plug-and-play optimal hyperparameter tool for the community. Its estimated values on the test set are merely 0.07\% away from the globally optimal LLM performance found via an exhaustive search. These laws demonstrate remarkable robustness across variations in model sparsity, training data distribution, and model shape. To our best known, this is the first work that unifies different model shapes and structures, such as Mixture-of-Experts models and dense transformers, as well as establishes optimal hyperparameter scaling laws across diverse data distributions. This exhaustive optimization process demands substantial computational resources, utilizing nearly one million NVIDIA H800 GPU hours to train 3,700 LLMs of varying sizes and hyperparameters from scratch and consuming approximately 100 trillion tokens in total. To facilitate reproducibility and further research, we will progressively release all loss measurements and model checkpoints through our designated repository https://step-law.github.io/
comment: 19 pages
☆ Scaling Rich Style-Prompted Text-to-Speech Datasets
We introduce Paralinguistic Speech Captions (ParaSpeechCaps), a large-scale dataset that annotates speech utterances with rich style captions. While rich abstract tags (e.g. guttural, nasal, pained) have been explored in small-scale human-annotated datasets, existing large-scale datasets only cover basic tags (e.g. low-pitched, slow, loud). We combine off-the-shelf text and speech embedders, classifiers and an audio language model to automatically scale rich tag annotations for the first time. ParaSpeechCaps covers a total of 59 style tags, including both speaker-level intrinsic tags and utterance-level situational tags. It consists of 342 hours of human-labelled data (PSC-Base) and 2427 hours of automatically annotated data (PSC-Scaled). We finetune Parler-TTS, an open-source style-prompted TTS model, on ParaSpeechCaps, and achieve improved style consistency (+7.9% Consistency MOS) and speech quality (+15.5% Naturalness MOS) over the best performing baseline that combines existing rich style tag datasets. We ablate several of our dataset design choices to lay the foundation for future work in this space. Our dataset, models and code are released at https://github.com/ajd12342/paraspeechcaps .
☆ Efficiently Escaping Saddle Points under Generalized Smoothness via Self-Bounding Regularity
In this paper, we study the problem of non-convex optimization on functions that are not necessarily smooth using first order methods. Smoothness (functions whose gradient and/or Hessian are Lipschitz) is not satisfied by many machine learning problems in both theory and practice, motivating a recent line of work studying the convergence of first order methods to first order stationary points under appropriate generalizations of smoothness. We develop a novel framework to study convergence of first order methods to first and \textit{second} order stationary points under generalized smoothness, under more general smoothness assumptions than the literature. Using our framework, we show appropriate variants of GD and SGD (e.g. with appropriate perturbations) can converge not just to first order but also \textit{second order stationary points} in runtime polylogarithmic in the dimension. To our knowledge, our work contains the first such result, as well as the first 'non-textbook' rate for non-convex optimization under generalized smoothness. We demonstrate that several canonical non-convex optimization problems fall under our setting and framework.
comment: 79 pages
☆ Sample-Optimal Agnostic Boosting with Unlabeled Data
Boosting provides a practical and provably effective framework for constructing accurate learning algorithms from inaccurate rules of thumb. It extends the promise of sample-efficient learning to settings where direct Empirical Risk Minimization (ERM) may not be implementable efficiently. In the realizable setting, boosting is known to offer this computational reprieve without compromising on sample efficiency. However, in the agnostic case, existing boosting algorithms fall short of achieving the optimal sample complexity. This paper highlights an unexpected and previously unexplored avenue of improvement: unlabeled samples. We design a computationally efficient agnostic boosting algorithm that matches the sample complexity of ERM, given polynomially many additional unlabeled samples. In fact, we show that the total number of samples needed, unlabeled and labeled inclusive, is never more than that for the best known agnostic boosting algorithm -- so this result is never worse -- while only a vanishing fraction of these need to be labeled for the algorithm to succeed. This is particularly fortuitous for learning-theoretic applications of agnostic boosting, which often take place in the distribution-specific setting, where unlabeled samples can be availed for free. We detail other applications of this result in reinforcement learning.
☆ Universality of Layer-Level Entropy-Weighted Quantization Beyond Model Architecture and Size
We present a novel approach to selective model quantization that transcends the limitations of architecture-specific and size-dependent compression methods for Large Language Models (LLMs) using Entropy-Weighted Quantization (EWQ). By analyzing the entropy distribution across transformer blocks, EWQ determines which blocks can be safely quantized without causing significant performance degradation, independent of model architecture or size. Our method outperforms uniform quantization approaches, maintaining Massive Multitask Language Understanding (MMLU) accuracy scores within 0.5% of unquantized models while reducing memory usage by up to 18%. We demonstrate the effectiveness of EWQ across multiple architectures-from 1.6B to 70B parameters-showcasing consistent improvements in the quality-compression trade-off regardless of model scale or architectural design. A surprising finding of EWQ is its ability to reduce perplexity compared to unquantized models, suggesting the presence of beneficial regularization through selective precision reduction. This improvement holds across different model families, indicating a fundamental relationship between layer-level entropy and optimal precision requirements. Additionally, we introduce FastEWQ, a rapid method for entropy distribution analysis that eliminates the need for loading model weights. This technique leverages universal characteristics of entropy distribution that persist across various architectures and scales, enabling near-instantaneous quantization decisions while maintaining 80% classification accuracy with full entropy analysis. Our results demonstrate that effective quantization strategies can be developed independently of specific architectural choices or model sizes, opening new possibilities for efficient LLM deployment.
comment: 29 pages, 7 figures, 14 tables; Comments are welcome
☆ L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning
Reasoning language models have shown an uncanny ability to improve performance at test-time by ``thinking longer''-that is, by generating longer chain-of-thought sequences and hence using more compute. However, the length of their chain-of-thought reasoning is not controllable, making it impossible to allocate test-time compute to achieve a desired level of performance. We introduce Length Controlled Policy Optimization (LCPO), a simple reinforcement learning method that optimizes for accuracy and adherence to user-specified length constraints. We use LCPO to train L1, a reasoning language model that produces outputs satisfying a length constraint given in its prompt. L1's length control allows for smoothly trading off computational cost and accuracy on a wide range of tasks, and outperforms the state-of-the-art S1 method for length control. Furthermore, we uncover an unexpected short chain-of-thought capability in models trained with LCPO. For instance, our 1.5B L1 model surpasses GPT-4o at equal reasoning lengths. Overall, LCPO enables precise control over reasoning length, allowing for fine-grained allocation of test-time compute and accuracy. We release code and models at https://www.cmu-l3.github.io/l1
☆ Coarse graining and reduced order models for plume ejection dynamics
Monitoring the atmospheric dispersion of pollutants is increasingly critical for environmental impact assessments. High-fidelity computational models are often employed to simulate plume dynamics, guiding decision-making and prioritizing resource deployment. However, such models can be prohibitively expensive to simulate, as they require resolving turbulent flows at fine spatial and temporal resolutions. Moreover, there are at least two distinct dynamical regimes of interest in the plume: (i) the initial ejection of the plume where turbulent mixing is generated by the shear-driven Kelvin-Helmholtz instability, and (ii) the ensuing turbulent diffusion and advection which is often modeled by the Gaussian plume model. We address the challenge of modeling the initial plume generation. Specifically, we propose a data-driven framework that identifies a reduced-order analytical model for plume dynamics -- directly from video data. We extract a time series of plume center and edge points from video snapshots and evaluate different regressions based to their extrapolation performance to generate a time series of coefficients that characterize the plume's overall direction and spread. We regress to a sinusoidal model inspired by the Kelvin-Helmholtz instability for the edge points in order to identify the plume's dispersion and vorticity. Overall, this reduced-order modeling framework provides a data-driven and lightweight approach to capture the dominant features of the initial nonlinear point-source plume dynamics, agnostic to plume type and starting only from video. The resulting model is a pre-cursor to standard models such as the Gaussian plume model and has the potential to enable rapid assessment and evaluation of critical environmental hazards, such as methane leaks, chemical spills, and pollutant dispersal from smokestacks.
☆ Compositional World Knowledge leads to High Utility Synthetic data
Machine learning systems struggle with robustness, under subpopulation shifts. This problem becomes especially pronounced in scenarios where only a subset of attribute combinations is observed during training -a severe form of subpopulation shift, referred as compositional shift. To address this problem, we ask the following question: Can we improve the robustness by training on synthetic data, spanning all possible attribute combinations? We first show that training of conditional diffusion models on limited data lead to incorrect underlying distribution. Therefore, synthetic data sampled from such models will result in unfaithful samples and does not lead to improve performance of downstream machine learning systems. To address this problem, we propose CoInD to reflect the compositional nature of the world by enforcing conditional independence through minimizing Fisher's divergence between joint and marginal distributions. We demonstrate that synthetic data generated by CoInD is faithful and this translates to state-of-the-art worst-group accuracy on compositional shift tasks on CelebA.
☆ Propagating Model Uncertainty through Filtering-based Probabilistic Numerical ODE Solvers
Filtering-based probabilistic numerical solvers for ordinary differential equations (ODEs), also known as ODE filters, have been established as efficient methods for quantifying numerical uncertainty in the solution of ODEs. In practical applications, however, the underlying dynamical system often contains uncertain parameters, requiring the propagation of this model uncertainty to the ODE solution. In this paper, we demonstrate that ODE filters, despite their probabilistic nature, do not automatically solve this uncertainty propagation problem. To address this limitation, we present a novel approach that combines ODE filters with numerical quadrature to properly marginalize over uncertain parameters, while accounting for both parameter uncertainty and numerical solver uncertainty. Experiments across multiple dynamical systems demonstrate that the resulting uncertainty estimates closely match reference solutions. Notably, we show how the numerical uncertainty from the ODE solver can help prevent overconfidence in the propagated uncertainty estimates, especially when using larger step sizes. Our results illustrate that probabilistic numerical methods can effectively quantify both numerical and parametric uncertainty in dynamical systems.
☆ Matrix Factorization for Inferring Associations and Missing Links
Missing link prediction is a method for network analysis, with applications in recommender systems, biology, social sciences, cybersecurity, information retrieval, and Artificial Intelligence (AI) reasoning in Knowledge Graphs. Missing link prediction identifies unseen but potentially existing connections in a network by analyzing the observed patterns and relationships. In proliferation detection, this supports efforts to identify and characterize attempts by state and non-state actors to acquire nuclear weapons or associated technology - a notoriously challenging but vital mission for global security. Dimensionality reduction techniques like Non-Negative Matrix Factorization (NMF) and Logistic Matrix Factorization (LMF) are effective but require selection of the matrix rank parameter, that is, of the number of hidden features, k, to avoid over/under-fitting. We introduce novel Weighted (WNMFk), Boolean (BNMFk), and Recommender (RNMFk) matrix factorization methods, along with ensemble variants incorporating logistic factorization, for link prediction. Our methods integrate automatic model determination for rank estimation by evaluating stability and accuracy using a modified bootstrap methodology and uncertainty quantification (UQ), assessing prediction reliability under random perturbations. We incorporate Otsu threshold selection and k-means clustering for Boolean matrix factorization, comparing them to coordinate descent-based Boolean thresholding. Our experiments highlight the impact of rank k selection, evaluate model performance under varying test-set sizes, and demonstrate the benefits of UQ for reliable predictions using abstention. We validate our methods on three synthetic datasets (Boolean and uniformly distributed) and benchmark them against LMF and symmetric LMF (symLMF) on five real-world protein-protein interaction networks, showcasing an improved prediction performance.
comment: 35 pages, 14 figures, 3 tables, 1 algorithm
☆ Multi-Agent Inverse Q-Learning from Demonstrations ICRA
When reward functions are hand-designed, deep reinforcement learning algorithms often suffer from reward misspecification, causing them to learn suboptimal policies in terms of the intended task objectives. In the single-agent case, inverse reinforcement learning (IRL) techniques attempt to address this issue by inferring the reward function from expert demonstrations. However, in multi-agent problems, misalignment between the learned and true objectives is exacerbated due to increased environment non-stationarity and variance that scales with multiple agents. As such, in multi-agent general-sum games, multi-agent IRL algorithms have difficulty balancing cooperative and competitive objectives. To address these issues, we propose Multi-Agent Marginal Q-Learning from Demonstrations (MAMQL), a novel sample-efficient framework for multi-agent IRL. For each agent, MAMQL learns a critic marginalized over the other agents' policies, allowing for a well-motivated use of Boltzmann policies in the multi-agent context. We identify a connection between optimal marginalized critics and single-agent soft-Q IRL, allowing us to apply a direct, simple optimization criterion from the single-agent domain. Across our experiments on three different simulated domains, MAMQL significantly outperforms previous multi-agent methods in average reward, sample efficiency, and reward recovery by often more than 2-5x. We make our code available at https://sites.google.com/view/mamql .
comment: 8 pages, 4 figures, 2 tables. Published at the International Conference on Robotics and Automation (ICRA) 2025
☆ An Information-theoretic Multi-task Representation Learning Framework for Natural Language Understanding AAAI 2025
This paper proposes a new principled multi-task representation learning framework (InfoMTL) to extract noise-invariant sufficient representations for all tasks. It ensures sufficiency of shared representations for all tasks and mitigates the negative effect of redundant features, which can enhance language understanding of pre-trained language models (PLMs) under the multi-task paradigm. Firstly, a shared information maximization principle is proposed to learn more sufficient shared representations for all target tasks. It can avoid the insufficiency issue arising from representation compression in the multi-task paradigm. Secondly, a task-specific information minimization principle is designed to mitigate the negative effect of potential redundant features in the input for each task. It can compress task-irrelevant redundant information and preserve necessary information relevant to the target for multi-task prediction. Experiments on six classification benchmarks show that our method outperforms 12 comparative multi-task methods under the same multi-task settings, especially in data-constrained and noisy scenarios. Extensive experiments demonstrate that the learned representations are more sufficient, data-efficient, and robust.
comment: 11 pages, accepted to AAAI 2025 (main conference), the code is available at https://github.com/zerohd4869/InfoMTL
☆ CLDyB: Towards Dynamic Benchmarking for Continual Learning with Pre-trained Models
The advent of the foundation model era has sparked significant research interest in leveraging pre-trained representations for continual learning (CL), yielding a series of top-performing CL methods on standard evaluation benchmarks. Nonetheless, there are growing concerns regarding potential data contamination during the pre-training stage. Furthermore, standard evaluation benchmarks, which are typically static, fail to capture the complexities of real-world CL scenarios, resulting in saturated performance. To address these issues, we describe CL on dynamic benchmarks (CLDyB), a general computational framework based on Markov decision processes for evaluating CL methods reliably. CLDyB dynamically identifies inherently difficult and algorithm-dependent tasks for the given CL methods, and determines challenging task orders using Monte Carlo tree search. Leveraging CLDyB, we first conduct a joint evaluation of multiple state-of-the-art CL methods, leading to a set of commonly challenging and generalizable task sequences where existing CL methods tend to perform poorly. We then conduct separate evaluations of individual CL methods using CLDyB, discovering their respective strengths and weaknesses. The source code and generated task sequences are publicly accessible at https://github.com/szc12153/CLDyB.
☆ Joint Masked Reconstruction and Contrastive Learning for Mining Interactions Between Proteins
Protein-protein interaction (PPI) prediction is an instrumental means in elucidating the mechanisms underlying cellular operations, holding significant practical implications for the realms of pharmaceutical development and clinical treatment. Presently, the majority of research methods primarily concentrate on the analysis of amino acid sequences, while investigations predicated on protein structures remain in the nascent stages of exploration. Despite the emergence of several structure-based algorithms in recent years, these are still confronted with inherent challenges: (1) the extraction of intrinsic structural information of proteins typically necessitates the expenditure of substantial computational resources; (2) these models are overly reliant on seen protein data, struggling to effectively unearth interaction cues between unknown proteins. To further propel advancements in this domain, this paper introduces a novel PPI prediction method jointing masked reconstruction and contrastive learning, termed JmcPPI. This methodology dissects the PPI prediction task into two distinct phases: during the residue structure encoding phase, JmcPPI devises two feature reconstruction tasks and employs graph attention mechanism to capture structural information between residues; during the protein interaction inference phase, JmcPPI perturbs the original PPI graph and employs a multi-graph contrastive learning strategy to thoroughly mine extrinsic interaction information of novel proteins. Extensive experiments conducted on three widely utilized PPI datasets demonstrate that JmcPPI surpasses existing optimal baseline models across various data partition schemes. The associated code can be accessed via https://github.com/lijfrank-open/JmcPPI.
comment: Submitted
☆ Transferable Foundation Models for Geometric Tasks on Point Cloud Representations: Geometric Neural Operators
We introduce methods for obtaining pretrained Geometric Neural Operators (GNPs) that can serve as basal foundation models for use in obtaining geometric features. These can be used within data processing pipelines for machine learning tasks and numerical methods. We show how our GNPs can be trained to learn robust latent representations for the differential geometry of point-clouds to provide estimates of metric, curvature, and other shape-related features. We demonstrate how our pre-trained GNPs can be used (i) to estimate the geometric properties of surfaces of arbitrary shape and topologies with robustness in the presence of noise, (ii) to approximate solutions of geometric partial differential equations (PDEs) on manifolds, and (iii) to solve equations for shape deformations such as curvature driven flows. We also release a package of the codes and weights for using our pre-trained GNPs for processing point cloud representations. This allows for incorporating our pre-trained GNPs as components for reuse within existing and new data processing pipelines. The GNPs also can be used as part of numerical solvers involving geometry or as part of methods for performing inference and other geometric tasks.
☆ Simulating the Real World: A Unified Survey of Multimodal Generative Models
Understanding and replicating the real world is a critical challenge in Artificial General Intelligence (AGI) research. To achieve this, many existing approaches, such as world models, aim to capture the fundamental principles governing the physical world, enabling more accurate simulations and meaningful interactions. However, current methods often treat different modalities, including 2D (images), videos, 3D, and 4D representations, as independent domains, overlooking their interdependencies. Additionally, these methods typically focus on isolated dimensions of reality without systematically integrating their connections. In this survey, we present a unified survey for multimodal generative models that investigate the progression of data dimensionality in real-world simulation. Specifically, this survey starts from 2D generation (appearance), then moves to video (appearance+dynamics) and 3D generation (appearance+geometry), and finally culminates in 4D generation that integrate all dimensions. To the best of our knowledge, this is the first attempt to systematically unify the study of 2D, video, 3D and 4D generation within a single framework. To guide future research, we provide a comprehensive review of datasets, evaluation metrics and future directions, and fostering insights for newcomers. This survey serves as a bridge to advance the study of multimodal generative models and real-world simulation within a unified framework.
comment: Repository for the related papers at https://github.com/ALEEEHU/World-Simulator
☆ Enhancing SAM with Efficient Prompting and Preference Optimization for Semi-supervised Medical Image Segmentation CVPR 2025
Foundational models such as the Segment Anything Model (SAM) are gaining traction in medical imaging segmentation, supporting multiple downstream tasks. However, such models are supervised in nature, still relying on large annotated datasets or prompts supplied by experts. Conventional techniques such as active learning to alleviate such limitations are limited in scope and still necessitate continuous human involvement and complex domain knowledge for label refinement or establishing reward ground truth. To address these challenges, we propose an enhanced Segment Anything Model (SAM) framework that utilizes annotation-efficient prompts generated in a fully unsupervised fashion, while still capturing essential semantic, location, and shape information through contrastive language-image pretraining and visual question answering. We adopt the direct preference optimization technique to design an optimal policy that enables the model to generate high-fidelity segmentations with simple ratings or rankings provided by a virtual annotator simulating the human annotation process. State-of-the-art performance of our framework in tasks such as lung segmentation, breast tumor segmentation, and organ segmentation across various modalities, including X-ray, ultrasound, and abdominal CT, justifies its effectiveness in low-annotation data scenarios.
comment: Accepted to CVPR 2025
☆ No Forgetting Learning: Memory-free Continual Learning ICCV 2025
Continual Learning (CL) remains a central challenge in deep learning, where models must sequentially acquire new knowledge while mitigating Catastrophic Forgetting (CF) of prior tasks. Existing approaches often struggle with efficiency and scalability, requiring extensive memory or model buffers. This work introduces ``No Forgetting Learning" (NFL), a memory-free CL framework that leverages knowledge distillation to maintain stability while preserving plasticity. Memory-free means the NFL does not rely on any memory buffer. Through extensive evaluations of three benchmark datasets, we demonstrate that NFL achieves competitive performance while utilizing approximately 14.75 times less memory than state-of-the-art methods. Furthermore, we introduce a new metric to better assess CL's plasticity-stability trade-off.
comment: This paper is submitted to ICCV 2025
☆ Mark Your LLM: Detecting the Misuse of Open-Source Large Language Models via Watermarking ICLR 2025
As open-source large language models (LLMs) like Llama3 become more capable, it is crucial to develop watermarking techniques to detect their potential misuse. Existing watermarking methods either add watermarks during LLM inference, which is unsuitable for open-source LLMs, or primarily target classification LLMs rather than recent generative LLMs. Adapting these watermarks to open-source LLMs for misuse detection remains an open challenge. This work defines two misuse scenarios for open-source LLMs: intellectual property (IP) violation and LLM Usage Violation. Then, we explore the application of inference-time watermark distillation and backdoor watermarking in these contexts. We propose comprehensive evaluation methods to assess the impact of various real-world further fine-tuning scenarios on watermarks and the effect of these watermarks on LLM performance. Our experiments reveal that backdoor watermarking could effectively detect IP Violation, while inference-time watermark distillation is applicable in both scenarios but less robust to further fine-tuning and has a more significant impact on LLM performance compared to backdoor watermarking. Exploring more advanced watermarking methods for open-source LLMs to detect their misuse should be an important future direction.
comment: Accepted by the 1st Workshop on GenAI Watermarking, collocated with ICLR 2025
☆ IDInit: A Universal and Stable Initialization Method for Neural Network Training ICLR 2025
Deep neural networks have achieved remarkable accomplishments in practice. The success of these networks hinges on effective initialization methods, which are vital for ensuring stable and rapid convergence during training. Recently, initialization methods that maintain identity transition within layers have shown good efficiency in network training. These techniques (e.g., Fixup) set specific weights to zero to achieve identity control. However, settings of remaining weight (e.g., Fixup uses random values to initialize non-zero weights) will affect the inductive bias that is achieved only by a zero weight, which may be harmful to training. Addressing this concern, we introduce fully identical initialization (IDInit), a novel method that preserves identity in both the main and sub-stem layers of residual networks. IDInit employs a padded identity-like matrix to overcome rank constraints in non-square weight matrices. Furthermore, we show the convergence problem of an identity matrix can be solved by stochastic gradient descent. Additionally, we enhance the universality of IDInit by processing higher-order weights and addressing dead neuron problems. IDInit is a straightforward yet effective initialization method, with improved convergence, stability, and performance across various settings, including large-scale datasets and deep models.
comment: Accepted in ICLR 2025
☆ The Best of Both Worlds: Integrating Language Models and Diffusion Models for Video Generation
Recent advancements in text-to-video (T2V) generation have been driven by two competing paradigms: autoregressive language models and diffusion models. However, each paradigm has intrinsic limitations: language models struggle with visual quality and error accumulation, while diffusion models lack semantic understanding and causal modeling. In this work, we propose LanDiff, a hybrid framework that synergizes the strengths of both paradigms through coarse-to-fine generation. Our architecture introduces three key innovations: (1) a semantic tokenizer that compresses 3D visual features into compact 1D discrete representations through efficient semantic compression, achieving a $\sim$14,000$\times$ compression ratio; (2) a language model that generates semantic tokens with high-level semantic relationships; (3) a streaming diffusion model that refines coarse semantics into high-fidelity videos. Experiments show that LanDiff, a 5B model, achieves a score of 85.43 on the VBench T2V benchmark, surpassing the state-of-the-art open-source models Hunyuan Video (13B) and other commercial models such as Sora, Keling, and Hailuo. Furthermore, our model also achieves state-of-the-art performance in long video generation, surpassing other open-source models in this field. Our demo can be viewed at https://landiff.github.io/.
☆ Fusion of Various Optimization Based Feature Smoothing Methods for Wearable and Non-invasive Blood Glucose Estimation
Recently, the wearable and non-invasive blood glucose estimation approach has been proposed. However, due to the unreliability of the acquisition device, the presence of the noise and the variations of the acquisition environments, the obtained features and the reference blood glucose values are highly unreliable. To address this issue, this paper proposes a polynomial fitting approach to smooth the obtained features or the reference blood glucose values. First, the blood glucose values are estimated based on the individual optimization approaches. Second, the absolute difference values between the estimated blood glucose values and the actual blood glucose values based on each optimization approach are computed. Third, these absolute difference values for each optimization approach are sorted in the ascending order. Fourth, for each sorted blood glucose value, the optimization method corresponding to the minimum absolute difference value is selected. Fifth, the accumulate probability of each selected optimization method is computed. If the accumulate probability of any selected optimization method at a point is greater than a threshold value, then the accumulate probabilities of these three selected optimization methods at that point are reset to zero. A range of the sorted blood glucose values are defined as that with the corresponding boundaries points being the previous reset point and this reset point. Hence, after performing the above procedures for all the sorted reference blood glucose values in the validation set, the regions of the sorted reference blood glucose values and the corresponding optimization methods in these regions are determined. The computer numerical simulation results show that our proposed method yields the mean absolute relative deviation (MARD) at 0.0930 and the percentage of the test data falling in the zone A of the Clarke error grid at 94.1176%.
comment: This version corrects several typos
☆ HybridNorm: Towards Stable and Efficient Transformer Training via Hybrid Normalization
Transformers have become the de facto architecture for a wide range of machine learning tasks, particularly in large language models (LLMs). Despite their remarkable performance, challenges remain in training deep transformer networks, especially regarding the location of layer normalization. While Pre-Norm structures facilitate easier training due to their more prominent identity path, they often yield suboptimal performance compared to Post-Norm. In this paper, we propose $\textbf{HybridNorm}$, a straightforward yet effective hybrid normalization strategy that integrates the advantages of both Pre-Norm and Post-Norm approaches. Specifically, HybridNorm employs QKV normalization within the attention mechanism and Post-Norm in the feed-forward network (FFN) of each transformer block. This design not only stabilizes training but also enhances performance, particularly in the context of LLMs. Comprehensive experiments in both dense and sparse architectures show that HybridNorm consistently outperforms both Pre-Norm and Post-Norm approaches, achieving state-of-the-art results across various benchmarks. These findings highlight the potential of HybridNorm as a more stable and effective technique for improving the training and performance of deep transformer models. %Code will be made publicly available. Code is available at https://github.com/BryceZhuo/HybridNorm.
☆ Advancing Solutions for the Three-Body Problem Through Physics-Informed Neural Networks
First formulated by Sir Isaac Newton in his work "Philosophiae Naturalis Principia Mathematica", the concept of the Three-Body Problem was put forth as a study of the motion of the three celestial bodies within the Earth-Sun-Moon system. In a generalized definition, it seeks to predict the motion for an isolated system composed of three point masses freely interacting under Newton's law of universal attraction. This proves to be analogous to a multitude of interactions between celestial bodies, and thus, the problem finds applicability within the studies of celestial mechanics. Despite numerous attempts by renowned physicists to solve it throughout the last three centuries, no general closed-form solutions have been reached due to its inherently chaotic nature for most initial conditions. Current state-of-the-art solutions are based on two approaches, either numerical high-precision integration or machine learning-based. Notwithstanding the breakthroughs of neural networks, these present a significant limitation, which is their ignorance of any prior knowledge of the chaotic systems presented. Thus, in this work, we propose a novel method that utilizes Physics-Informed Neural Networks (PINNs). These deep neural networks are able to incorporate any prior system knowledge expressible as an Ordinary Differential Equation (ODE) into their learning processes as a regularizing agent. Our findings showcase that PINNs surpass current state-of-the-art machine learning methods with comparable prediction quality. Despite a better prediction quality, the usability of numerical integrators suffers due to their prohibitively high computational cost. These findings confirm that PINNs are both effective and time-efficient open-form solvers of the Three-Body Problem that capitalize on the extensive knowledge we hold of classical mechanics.
comment: 14 pages, 25 figures, 3 tables. 75th International Astronautical Congress (IAC), Milan, Italy, 14-18 October
☆ PSDNorm: Test-Time Temporal Normalization for Deep Learning on EEG Signals
Distribution shift poses a significant challenge in machine learning, particularly in biomedical applications such as EEG signals collected across different subjects, institutions, and recording devices. While existing normalization layers, Batch-Norm, LayerNorm and InstanceNorm, help address distribution shifts, they fail to capture the temporal dependencies inherent in temporal signals. In this paper, we propose PSDNorm, a layer that leverages Monge mapping and temporal context to normalize feature maps in deep learning models. Notably, the proposed method operates as a test-time domain adaptation technique, addressing distribution shifts without additional training. Evaluations on 10 sleep staging datasets using the U-Time model demonstrate that PSDNorm achieves state-of-the-art performance at test time on datasets not seen during training while being 4x more data-efficient than the best baseline. Additionally, PSDNorm provides a significant improvement in robustness, achieving markedly higher F1 scores for the 20% hardest subjects.
☆ Data-augmented Learning of Geodesic Distances in Irregular Domains through Soner Boundary Conditions
Geodesic distances play a fundamental role in robotics, as they efficiently encode global geometric information of the domain. Recent methods use neural networks to approximate geodesic distances by solving the Eikonal equation through physics-informed approaches. While effective, these approaches often suffer from unstable convergence during training in complex environments. We propose a framework to learn geodesic distances in irregular domains by using the Soner boundary condition, and systematically evaluate the impact of data losses on training stability and solution accuracy. Our experiments demonstrate that incorporating data losses significantly improves convergence robustness, reducing training instabilities and sensitivity to initialization. These findings suggest that hybrid data-physics approaches can effectively enhance the reliability of learning-based geodesic distance solvers with sparse data.
☆ Meta Learning not to Learn: Robustly Informing Meta-Learning under Nuisance-Varying Families
In settings where both spurious and causal predictors are available, standard neural networks trained under the objective of empirical risk minimization (ERM) with no additional inductive biases tend to have a dependence on a spurious feature. As a result, it is necessary to integrate additional inductive biases in order to guide the network toward generalizable hypotheses. Often these spurious features are shared across related tasks, such as estimating disease prognoses from image scans coming from different hospitals, making the challenge of generalization more difficult. In these settings, it is important that methods are able to integrate the proper inductive biases to generalize across both nuisance-varying families as well as task families. Motivated by this setting, we present RIME (Robustly Informed Meta lEarning), a new method for meta learning under the presence of both positive and negative inductive biases (what to learn and what not to learn). We first develop a theoretical causal framework showing why existing approaches at knowledge integration can lead to worse performance on distributionally robust objectives. We then show that RIME is able to simultaneously integrate both biases, reaching state of the art performance under distributionally robust objectives in informed meta-learning settings under nuisance-varying families.
☆ Compositional Causal Reasoning Evaluation in Language Models
Causal reasoning and compositional reasoning are two core aspirations in generative AI. Measuring the extent of these behaviors requires principled evaluation methods. We explore a unified perspective that considers both behaviors simultaneously, termed compositional causal reasoning (CCR): the ability to infer how causal measures compose and, equivalently, how causal quantities propagate through graphs. We instantiate a framework for the systematic evaluation of CCR for the average treatment effect and the probability of necessity and sufficiency. As proof of concept, we demonstrate the design of CCR tasks for language models in the LLama, Phi, and GPT families. On a math word problem, our framework revealed a range of taxonomically distinct error patterns. Additionally, CCR errors increased with the complexity of causal paths for all models except o1.
☆ Federated Dynamic Modeling and Learning for Spatiotemporal Data Forecasting
This paper presents an advanced Federated Learning (FL) framework for forecasting complex spatiotemporal data, improving upon recent state-of-the-art models. In the proposed approach, the original Gated Recurrent Unit (GRU) module within previous Dynamic Spatial--Temporal Graph Convolutional Recurrent Network (DSTGCRN) modeling is first replaced with a Long Short-Term Memory (LSTM) network, enabling the resulting model to more effectively capture long-term dependencies inherent to time series data. The resulting architecture significantly improves the model's capacity to handle complex temporal patterns in diverse forecasting applications. Furthermore, the proposed FL framework integrates a novel Client-Side Validation (CSV) mechanism, introducing a critical validation step at the client level before incorporating aggregated parameters from the central server into local models. This ensures that only the most effective updates are adopted, improving both the robustness and accuracy of the forecasting model across clients. The efficiency of our approach is demonstrated through extensive experiments on real-world applications, including public datasets for multimodal transport demand forecasting and private datasets for Origin-Destination (OD) matrix forecasting in urban areas. The results demonstrate substantial improvements over conventional methods, highlighting the framework's ability to capture complex spatiotemporal dependencies while preserving data privacy. This work not only provides a scalable and privacy-preserving solution for real-time, region-specific forecasting and management but also underscores the potential of leveraging distributed data sources in a FL context. We provide our algorithms as open-source on GitHub.
☆ Leveraging priors on distribution functions for multi-arm bandits
We introduce Dirichlet Process Posterior Sampling (DPPS), a Bayesian non-parametric algorithm for multi-arm bandits based on Dirichlet Process (DP) priors. Like Thompson-sampling, DPPS is a probability-matching algorithm, i.e., it plays an arm based on its posterior-probability of being optimal. Instead of assuming a parametric class for the reward generating distribution of each arm, and then putting a prior on the parameters, in DPPS the reward generating distribution is directly modeled using DP priors. DPPS provides a principled approach to incorporate prior belief about the bandit environment, and in the noninformative limit of the DP posteriors (i.e. Bayesian Bootstrap), we recover Non Parametric Thompson Sampling (NPTS), a popular non-parametric bandit algorithm, as a special case of DPPS. We employ stick-breaking representation of the DP priors, and show excellent empirical performance of DPPS in challenging synthetic and real world bandit environments. Finally, using an information-theoretic analysis, we show non-asymptotic optimality of DPPS in the Bayesian regret setup.
☆ STX-Search: Explanation Search for Continuous Dynamic Spatio-Temporal Models
Recent improvements in the expressive power of spatio-temporal models have led to performance gains in many real-world applications, such as traffic forecasting and social network modelling. However, understanding the predictions from a model is crucial to ensure reliability and trustworthiness, particularly for high-risk applications, such as healthcare and transport. Few existing methods are able to generate explanations for models trained on continuous-time dynamic graph data and, of these, the computational complexity and lack of suitable explanation objectives pose challenges. In this paper, we propose $\textbf{S}$patio-$\textbf{T}$emporal E$\textbf{X}$planation $\textbf{Search}$ (STX-Search), a novel method for generating instance-level explanations that is applicable to static and dynamic temporal graph structures. We introduce a novel search strategy and objective function, to find explanations that are highly faithful and interpretable. When compared with existing methods, STX-Search produces explanations of higher fidelity whilst optimising explanation size to maintain interpretability.
☆ A Morse Transform for Drug Discovery
We introduce a new ligand-based virtual screening (LBVS) framework that uses piecewise linear (PL) Morse theory to predict ligand binding potential. We model ligands as simplicial complexes via a pruned Delaunay triangulation, and catalogue the critical points across multiple directional height functions. This produces a rich feature vector, consisting of crucial topological features -- peaks, troughs, and saddles -- that characterise ligand surfaces relevant to binding interactions. Unlike contemporary LBVS methods that rely on computationally-intensive deep neural networks, we require only a lightweight classifier. The Morse theoretic approach achieves state-of-the-art performance on standard datasets while offering an interpretable feature vector and scalable method for ligand prioritization in early-stage drug discovery.
comment: 25 pages, 5 main figures, 2 main tables, 6 supplementary figures and 4 supplementary tables
☆ Learning Object Placement Programs for Indoor Scene Synthesis with Iterative Self Training
Data driven and autoregressive indoor scene synthesis systems generate indoor scenes automatically by suggesting and then placing objects one at a time. Empirical observations show that current systems tend to produce incomplete next object location distributions. We introduce a system which addresses this problem. We design a Domain Specific Language (DSL) that specifies functional constraints. Programs from our language take as input a partial scene and object to place. Upon execution they predict possible object placements. We design a generative model which writes these programs automatically. Available 3D scene datasets do not contain programs to train on, so we build upon previous work in unsupervised program induction to introduce a new program bootstrapping algorithm. In order to quantify our empirical observations we introduce a new evaluation procedure which captures how well a system models per-object location distributions. We ask human annotators to label all the possible places an object can go in a scene and show that our system produces per-object location distributions more consistent with human annotators. Our system also generates indoor scenes of comparable quality to previous systems and while previous systems degrade in performance when training data is sparse, our system does not degrade to the same degree.
comment: 21 pages, 20 figures Subjects: Graphics (cs.GR), Computer Vision and Pattern Recognition (cs.CV), Machine Learning (cs.LG)
☆ Accurate predictive model of band gap with selected important features based on explainable machine learning
In the rapidly advancing field of materials informatics, nonlinear machine learning models have demonstrated exceptional predictive capabilities for material properties. However, their black-box nature limits interpretability, and they may incorporate features that do not contribute to, or even deteriorate, model performance. This study employs explainable ML (XML) techniques, including permutation feature importance and the SHapley Additive exPlanation, applied to a pristine support vector regression model designed to predict band gaps at the GW level using 18 input features. Guided by XML-derived individual feature importance, a simple framework is proposed to construct reduced-feature predictive models. Model evaluations indicate that an XML-guided compact model, consisting of the top five features, achieves comparable accuracy to the pristine model on in-domain datasets while demonstrating superior generalization with lower prediction errors on out-of-domain data. Additionally, the study underscores the necessity for eliminating strongly correlated features to prevent misinterpretation and overestimation of feature importance before applying XML. This study highlights XML's effectiveness in developing simplified yet highly accurate machine learning models by clarifying feature roles.
comment: 9 pages, 4 figures, SI is included
☆ InfoSEM: A Deep Generative Model with Informative Priors for Gene Regulatory Network Inference ICLR 2025
Inferring Gene Regulatory Networks (GRNs) from gene expression data is crucial for understanding biological processes. While supervised models are reported to achieve high performance for this task, they rely on costly ground truth (GT) labels and risk learning gene-specific biases, such as class imbalances of GT interactions, rather than true regulatory mechanisms. To address these issues, we introduce InfoSEM, an unsupervised generative model that leverages textual gene embeddings as informative priors, improving GRN inference without GT labels. InfoSEM can also integrate GT labels as an additional prior when available, avoiding biases and further enhancing performance. Additionally, we propose a biologically motivated benchmarking framework that better reflects real-world applications such as biomarker discovery and reveals learned biases of existing supervised methods. InfoSEM outperforms existing models by 38.5% across four datasets using textual embeddings prior and further boosts performance by 11.1% when integrating labeled data as priors.
comment: ICLR 2025 AI4NA Oral, ICLR 2025 MLGenX Spotlight, ICLR 2025 LMRL
☆ Generalized Interpolating Discrete Diffusion
While state-of-the-art language models achieve impressive results through next-token prediction, they have inherent limitations such as the inability to revise already generated tokens. This has prompted exploration of alternative approaches such as discrete diffusion. However, masked diffusion, which has emerged as a popular choice due to its simplicity and effectiveness, reintroduces this inability to revise words. To overcome this, we generalize masked diffusion and derive the theoretical backbone of a family of general interpolating discrete diffusion (GIDD) processes offering greater flexibility in the design of the noising processes. Leveraging a novel diffusion ELBO, we achieve compute-matched state-of-the-art performance in diffusion language modeling. Exploiting GIDD's flexibility, we explore a hybrid approach combining masking and uniform noise, leading to improved sample quality and unlocking the ability for the model to correct its own mistakes, an area where autoregressive models notoriously have struggled. Our code and models are open-source: https://github.com/dvruette/gidd/
☆ Poisoning Bayesian Inference via Data Deletion and Replication
Research in adversarial machine learning (AML) has shown that statistical models are vulnerable to maliciously altered data. However, despite advances in Bayesian machine learning models, most AML research remains concentrated on classical techniques. Therefore, we focus on extending the white-box model poisoning paradigm to attack generic Bayesian inference, highlighting its vulnerability in adversarial contexts. A suite of attacks are developed that allow an attacker to steer the Bayesian posterior toward a target distribution through the strategic deletion and replication of true observations, even when only sampling access to the posterior is available. Analytic properties of these algorithms are proven and their performance is empirically examined in both synthetic and real-world scenarios. With relatively little effort, the attacker is able to substantively alter the Bayesian's beliefs and, by accepting more risk, they can mold these beliefs to their will. By carefully constructing the adversarial posterior, surgical poisoning is achieved such that only targeted inferences are corrupted and others are minimally disturbed.
☆ Know Thy Judge: On the Robustness Meta-Evaluation of LLM Safety Judges ICLR'25
Large Language Model (LLM) based judges form the underpinnings of key safety evaluation processes such as offline benchmarking, automated red-teaming, and online guardrailing. This widespread requirement raises the crucial question: can we trust the evaluations of these evaluators? In this paper, we highlight two critical challenges that are typically overlooked: (i) evaluations in the wild where factors like prompt sensitivity and distribution shifts can affect performance and (ii) adversarial attacks that target the judge. We highlight the importance of these through a study of commonly used safety judges, showing that small changes such as the style of the model output can lead to jumps of up to 0.24 in the false negative rate on the same dataset, whereas adversarial attacks on the model generation can fool some judges into misclassifying 100% of harmful generations as safe ones. These findings reveal gaps in commonly used meta-evaluation benchmarks and weaknesses in the robustness of current LLM judges, indicating that low attack success under certain judges could create a false sense of security.
comment: Accepted to the ICBINB Workshop at ICLR'25
☆ DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models
Recent advancements in slow-thinking reasoning models have shown exceptional performance in complex reasoning tasks. However, these models often exhibit overthinking-generating redundant reasoning steps for simple problems, leading to excessive computational resource usage. While current mitigation strategies uniformly reduce reasoning tokens, they risk degrading performance on challenging tasks that require extended reasoning. This paper introduces Difficulty-Adaptive Slow-Thinking (DAST), a novel framework that enables models to autonomously adjust the length of Chain-of-Thought(CoT) based on problem difficulty. We first propose a Token Length Budget (TLB) metric to quantify difficulty, then leveraging length-aware reward shaping and length preference optimization to implement DAST. DAST penalizes overlong responses for simple tasks while incentivizing sufficient reasoning for complex problems. Experiments on diverse datasets and model scales demonstrate that DAST effectively mitigates overthinking (reducing token usage by over 30\% on average) while preserving reasoning accuracy on complex problems.
comment: working in progress
☆ An artificially intelligent magnetic resonance spectroscopy quantification method: Comparison between QNet and LCModel on the cloud computing platform CloudBrain-MRS
Objctives: This work aimed to statistically compare the metabolite quantification of human brain magnetic resonance spectroscopy (MRS) between the deep learning method QNet and the classical method LCModel through an easy-to-use intelligent cloud computing platform CloudBrain-MRS. Materials and Methods: In this retrospective study, two 3 T MRI scanners Philips Ingenia and Achieva collected 61 and 46 in vivo 1H magnetic resonance (MR) spectra of healthy participants, respectively, from the brain region of pregenual anterior cingulate cortex from September to October 2021. The analyses of Bland-Altman, Pearson correlation and reasonability were performed to assess the degree of agreement, linear correlation and reasonability between the two quantification methods. Results: Fifteen healthy volunteers (12 females and 3 males, age range: 21-35 years, mean age/standard deviation = 27.4/3.9 years) were recruited. The analyses of Bland-Altman, Pearson correlation and reasonability showed high to good consistency and very strong to moderate correlation between the two methods for quantification of total N-acetylaspartate (tNAA), total choline (tCho), and inositol (Ins) (relative half interval of limits of agreement = 3.04%, 9.3%, and 18.5%, respectively; Pearson correlation coefficient r = 0.775, 0.927, and 0.469, respectively). In addition, quantification results of QNet are more likely to be closer to the previous reported average values than those of LCModel. Conclusion: There were high or good degrees of consistency between the quantification results of QNet and LCModel for tNAA, tCho, and Ins, and QNet generally has more reasonable quantification than LCModel.
☆ PALo: Learning Posture-Aware Locomotion for Quadruped Robots
With the rapid development of embodied intelligence, locomotion control of quadruped robots on complex terrains has become a research hotspot. Unlike traditional locomotion control approaches focusing solely on velocity tracking, we pursue to balance the agility and robustness of quadruped robots on diverse and complex terrains. To this end, we propose an end-to-end deep reinforcement learning framework for posture-aware locomotion named PALo, which manages to handle simultaneous linear and angular velocity tracking and real-time adjustments of body height, pitch, and roll angles. In PALo, the locomotion control problem is formulated as a partially observable Markov decision process, and an asymmetric actor-critic architecture is adopted to overcome the sim-to-real challenge. Further, by incorporating customized training curricula, PALo achieves agile posture-aware locomotion control in simulated environments and successfully transfers to real-world settings without fine-tuning, allowing real-time control of the quadruped robot's locomotion and body posture across challenging terrains. Through in-depth experimental analysis, we identify the key components of PALo that contribute to its performance, further validating the effectiveness of the proposed method. The results of this study provide new possibilities for the low-level locomotion control of quadruped robots in higher dimensional command spaces and lay the foundation for future research on upper-level modules for embodied intelligence.
☆ Reproducibility Assessment of Magnetic Resonance Spectroscopy of Pregenual Anterior Cingulate Cortex across Sessions and Vendors via the Cloud Computing Platform CloudBrain-MRS
Given the need to elucidate the mechanisms underlying illnesses and their treatment, as well as the lack of harmonization of acquisition and post-processing protocols among different magnetic resonance system vendors, this work is to determine if metabolite concentrations obtained from different sessions, machine models and even different vendors of 3 T scanners can be highly reproducible and be pooled for diagnostic analysis, which is very valuable for the research of rare diseases. Participants underwent magnetic resonance imaging (MRI) scanning once on two separate days within one week (one session per day, each session including two proton magnetic resonance spectroscopy (1H-MRS) scans with no more than a 5-minute interval between scans (no off-bed activity)) on each machine. were analyzed for reliability of within- and between- sessions using the coefficient of variation (CV) and intraclass correlation coefficient (ICC), and for reproducibility of across the machines using correlation coefficient. As for within- and between- session, all CV values for a group of all the first or second scans of a session, or for a session were almost below 20%, and most of the ICCs for metabolites range from moderate (0.4-0.59) to excellent (0.75-1), indicating high data reliability. When it comes to the reproducibility across the three scanners, all Pearson correlation coefficients across the three machines approached 1 with most around 0.9, and majority demonstrated statistical significance (P<0.01). Additionally, the intra-vendor reproducibility was greater than the inter-vendor ones.
☆ Privacy Preserving and Robust Aggregation for Cross-Silo Federated Learning in Non-IID Settings
Federated Averaging remains the most widely used aggregation strategy in federated learning due to its simplicity and scalability. However, its performance degrades significantly in non-IID data settings, where client distributions are highly imbalanced or skewed. Additionally, it relies on clients transmitting metadata, specifically the number of training samples, which introduces privacy risks and may conflict with regulatory frameworks like the European GDPR. In this paper, we propose a novel aggregation strategy that addresses these challenges by introducing class-aware gradient masking. Unlike traditional approaches, our method relies solely on gradient updates, eliminating the need for any additional client metadata, thereby enhancing privacy protection. Furthermore, our approach validates and dynamically weights client contributions based on class-specific importance, ensuring robustness against non-IID distributions, convergence prevention, and backdoor attacks. Extensive experiments on benchmark datasets demonstrate that our method not only outperforms FedAvg and other widely accepted aggregation strategies in non-IID settings but also preserves model integrity in adversarial scenarios. Our results establish the effectiveness of gradient masking as a practical and secure solution for federated learning.
☆ A Graph-Partitioning Based Continuous Optimization Approach to Semi-supervised Clustering Problems
Semi-supervised clustering is a basic problem in various applications. Most existing methods require knowledge of the ideal cluster number, which is often difficult to obtain in practice. Besides, satisfying the must-link constraints is another major challenge for these methods. In this work, we view the semi-supervised clustering task as a partitioning problem on a graph associated with the given dataset, where the similarity matrix includes a scaling parameter to reflect the must-link constraints. Utilizing a relaxation technique, we formulate the graph partitioning problem into a continuous optimization model that does not require the exact cluster number, but only an overestimate of it. We then propose a block coordinate descent algorithm to efficiently solve this model, and establish its convergence result. Based on the obtained solution, we can construct the clusters that theoretically meet the must-link constraints under mild assumptions. Furthermore, we verify the effectiveness and efficiency of our proposed method through comprehensive numerical experiments.
☆ FORTALESA: Fault-Tolerant Reconfigurable Systolic Array for DNN Inference
The emergence of Deep Neural Networks (DNNs) in mission- and safety-critical applications brings their reliability to the front. High performance demands of DNNs require the use of specialized hardware accelerators. Systolic array architecture is widely used in DNN accelerators due to its parallelism and regular structure. This work presents a run-time reconfigurable systolic array architecture with three execution modes and four implementation options. All four implementations are evaluated in terms of resource utilization, throughput, and fault tolerance improvement. The proposed architecture is used for reliability enhancement of DNN inference on systolic array through heterogeneous mapping of different network layers to different execution modes. The approach is supported by a novel reliability assessment method based on fault propagation analysis. It is used for the exploration of the appropriate execution mode-layer mapping for DNN inference. The proposed architecture efficiently protects registers and MAC units of systolic array PEs from transient and permanent faults. The reconfigurability feature enables a speedup of up to $3\times$, depending on layer vulnerability. Furthermore, it requires $6\times$ less resources compared to static redundancy and $2.5\times$ less resources compared to the previously proposed solution for transient faults.
comment: 11 pages, 15 figures
☆ Determinant Estimation under Memory Constraints and Neural Scaling Laws
Calculating or accurately estimating log-determinants of large positive semi-definite matrices is of fundamental importance in many machine learning tasks. While its cubic computational complexity can already be prohibitive, in modern applications, even storing the matrices themselves can pose a memory bottleneck. To address this, we derive a novel hierarchical algorithm based on block-wise computation of the LDL decomposition for large-scale log-determinant calculation in memory-constrained settings. In extreme cases where matrices are highly ill-conditioned, accurately computing the full matrix itself may be infeasible. This is particularly relevant when considering kernel matrices at scale, including the empirical Neural Tangent Kernel (NTK) of neural networks trained on large datasets. Under the assumption of neural scaling laws in the test error, we show that the ratio of pseudo-determinants satisfies a power-law relationship, allowing us to derive corresponding scaling laws. This enables accurate estimation of NTK log-determinants from a tiny fraction of the full dataset; in our experiments, this results in a $\sim$100,000$\times$ speedup with improved accuracy over competing approximations. Using these techniques, we successfully estimate log-determinants for dense matrices of extreme sizes, which were previously deemed intractable and inaccessible due to their enormous scale and computational demands.
☆ AOLO: Analysis and Optimization For Low-Carbon Oriented Wireless Large Language Model Services
Recent advancements in large language models (LLMs) have led to their widespread adoption and large-scale deployment across various domains. However, their environmental impact, particularly during inference, has become a growing concern due to their substantial energy consumption and carbon footprint. Existing research has focused on inference computation alone, overlooking the analysis and optimization of carbon footprint in network-aided LLM service systems. To address this gap, we propose AOLO, a framework for analysis and optimization for low-carbon oriented wireless LLM services. AOLO introduces a comprehensive carbon footprint model that quantifies greenhouse gas emissions across the entire LLM service chain, including computational inference and wireless communication. Furthermore, we formulate an optimization problem aimed at minimizing the overall carbon footprint, which is solved through joint optimization of inference outputs and transmit power under quality-of-experience and system performance constraints. To achieve this joint optimization, we leverage the energy efficiency of spiking neural networks (SNNs) by adopting SNN as the actor network and propose a low-carbon-oriented optimization algorithm, i.e., SNN-based deep reinforcement learning (SDRL). Comprehensive simulations demonstrate that SDRL algorithm significantly reduces overall carbon footprint, achieving an 18.77% reduction compared to the benchmark soft actor-critic, highlighting its potential for enabling more sustainable LLM inference services.
☆ Learning Transformer-based World Models with Contrastive Predictive Coding
The DreamerV3 algorithm recently obtained remarkable performance across diverse environment domains by learning an accurate world model based on Recurrent Neural Networks (RNNs). Following the success of model-based reinforcement learning algorithms and the rapid adoption of the Transformer architecture for its superior training efficiency and favorable scaling properties, recent works such as STORM have proposed replacing RNN-based world models with Transformer-based world models using masked self-attention. However, despite the improved training efficiency of these methods, their impact on performance remains limited compared to the Dreamer algorithm, struggling to learn competitive Transformer-based world models. In this work, we show that the next state prediction objective adopted in previous approaches is insufficient to fully exploit the representation capabilities of Transformers. We propose to extend world model predictions to longer time horizons by introducing TWISTER (Transformer-based World model wIth contraSTivE Representations), a world model using action-conditioned Contrastive Predictive Coding to learn high-level temporal feature representations and improve the agent performance. TWISTER achieves a human-normalized mean score of 162% on the Atari 100k benchmark, setting a new record among state-of-the-art methods that do not employ look-ahead search.
☆ Training-Free Graph Filtering via Multimodal Feature Refinement for Extremely Fast Multimodal Recommendation
Multimodal recommender systems improve the performance of canonical recommender systems with no item features by utilizing diverse content types such as text, images, and videos, while alleviating inherent sparsity of user-item interactions and accelerating user engagement. However, current neural network-based models often incur significant computational overhead due to the complex training process required to learn and integrate information from multiple modalities. To overcome this limitation, we propose MultiModal-Graph Filtering (MM-GF), a training-free method based on the notion of graph filtering (GF) for efficient and accurate multimodal recommendations. Specifically, MM-GF first constructs multiple similarity graphs through nontrivial multimodal feature refinement such as robust scaling and vector shifting by addressing the heterogeneous characteristics across modalities. Then, MM-GF optimally fuses multimodal information using linear low-pass filters across different modalities. Extensive experiments on real-world benchmark datasets demonstrate that MM-GF not only improves recommendation accuracy by up to 13.35% compared to the best competitor but also dramatically reduces computational costs by achieving the runtime of less than 10 seconds.
comment: 10 pages, 6 figures, 6 tables
☆ Temporal Analysis of NetFlow Datasets for Network Intrusion Detection Systems
This paper investigates the temporal analysis of NetFlow datasets for machine learning (ML)-based network intrusion detection systems (NIDS). Although many previous studies have highlighted the critical role of temporal features, such as inter-packet arrival time and flow length/duration, in NIDS, the currently available NetFlow datasets for NIDS lack these temporal features. This study addresses this gap by creating and making publicly available a set of NetFlow datasets that incorporate these temporal features [1]. With these temporal features, we provide a comprehensive temporal analysis of NetFlow datasets by examining the distribution of various features over time and presenting time-series representations of NetFlow features. This temporal analysis has not been previously provided in the existing literature. We also borrowed an idea from signal processing, time frequency analysis, and tested it to see how different the time frequency signal presentations (TFSPs) are for various attacks. The results indicate that many attacks have unique patterns, which could help ML models to identify them more easily.
☆ Speculative MoE: Communication Efficient Parallel MoE Inference with Speculative Token and Expert Pre-scheduling
MoE (Mixture of Experts) prevails as a neural architecture that can scale modern transformer-based LLMs (Large Language Models) to unprecedented scales. Nevertheless, large MoEs' great demands of computing power, memory capacity and memory bandwidth make scalable serving a fundamental challenge and efficient parallel inference has become a requisite to attain adequate throughput under latency constraints. DeepSpeed-MoE, one state-of-the-art MoE inference framework, adopts a 3D-parallel paradigm including EP (Expert Parallelism), TP (Tensor Parallel) and DP (Data Parallelism). However, our analysis shows DeepSpeed-MoE's inference efficiency is largely bottlenecked by EP, which is implemented with costly all-to-all collectives to route token activation. Our work aims to boost DeepSpeed-MoE by strategically reducing EP's communication overhead with a technique named Speculative MoE. Speculative MoE has two speculative parallelization schemes, speculative token shuffling and speculative expert grouping, which predict outstanding tokens' expert routing paths and pre-schedule tokens and experts across devices to losslessly trim EP's communication volume. Besides DeepSpeed-MoE, we also build Speculative MoE into a prevailing MoE inference engine SGLang. Experiments show Speculative MoE can significantly boost state-of-the-art MoE inference frameworks on fast homogeneous and slow heterogeneous interconnects.
☆ Time-varying Factor Augmented Vector Autoregression with Grouped Sparse Autoencoder
Recent economic events, including the global financial crisis and COVID-19 pandemic, have exposed limitations in linear Factor Augmented Vector Autoregressive (FAVAR) models for forecasting and structural analysis. Nonlinear dimension techniques, particularly autoencoders, have emerged as promising alternatives in a FAVAR framework, but challenges remain in identifiability, interpretability, and integration with traditional nonlinear time series methods. We address these challenges through two contributions. First, we introduce a Grouped Sparse autoencoder that employs the Spike-and-Slab Lasso prior, with parameters under this prior being shared across variables of the same economic category, thereby achieving semi-identifiability and enhancing model interpretability. Second, we incorporate time-varying parameters into the VAR component to better capture evolving economic dynamics. Our empirical application to the US economy demonstrates that the Grouped Sparse autoencoder produces more interpretable factors through its parsimonious structure; and its combination with time-varying parameter VAR shows superior performance in both point and density forecasting. Impulse response analysis reveals that monetary policy shocks during recessions generate more moderate responses with higher uncertainty compared to expansionary periods.
☆ Dedicated Feedback and Edit Models Empower Inference-Time Scaling for Open-Ended General-Domain Tasks
Inference-Time Scaling has been critical to the success of recent models such as OpenAI o1 and DeepSeek R1. However, many techniques used to train models for inference-time scaling require tasks to have answers that can be verified, limiting their application to domains such as math, coding and logical reasoning. We take inspiration from how humans make first attempts, ask for detailed feedback from others and make improvements based on such feedback across a wide spectrum of open-ended endeavors. To this end, we collect data for and train dedicated Feedback and Edit Models that are capable of performing inference-time scaling for open-ended general-domain tasks. In our setup, one model generates an initial response, which are given feedback by a second model, that are then used by a third model to edit the response. We show that performance on Arena Hard, a benchmark strongly predictive of Chatbot Arena Elo can be boosted by scaling the number of initial response drafts, effective feedback and edited responses. When scaled optimally, our setup based on 70B models from the Llama 3 family can reach SoTA performance on Arena Hard at 92.7 as of 5 Mar 2025, surpassing OpenAI o1-preview-2024-09-12 with 90.4 and DeepSeek R1 with 92.3.
comment: 22 pages, 2 figures
☆ How can representation dimension dominate structurally pruned LLMs? ICLR 2025
Pruning assumes a subnetwork exists in the original deep neural network, which can achieve comparative model performance with less computation than the original. However, it is unclear how the model performance varies with the different subnetwork extractions. In this paper, we choose the representation dimension (or embedding dimension, model dimension, the dimension of the residual stream in the relevant literature) as the entry point to this issue. We investigate the linear transformations in the LLM transformer blocks and consider a specific structured pruning approach, SliceGPT, to extract the subnetworks of different representation dimensions. We mechanistically analyse the activation flow during the model forward passes, and find the representation dimension dominates the linear transformations, model predictions, and, finally, the model performance. Explicit analytical relations are given to calculate the pruned model performance (perplexity and accuracy) without actual evaluation, and are empirically validated with Llama-3-8B-Instruct and Phi-3-mini-4k-Instruct.
comment: ICLR 2025 Workshop on Sparsity in LLMs (SLLM)
☆ FILM: Framework for Imbalanced Learning Machines based on a new unbiased performance measure and a new ensemble-based technique
This research addresses the challenges of handling unbalanced datasets for binary classification tasks. In such scenarios, standard evaluation metrics are often biased by the disproportionate representation of the minority class. Conducting experiments across seven datasets, we uncovered inconsistencies in evaluation metrics when determining the model that outperforms others for each binary classification problem. This justifies the need for a metric that provides a more consistent and unbiased evaluation across unbalanced datasets, thereby supporting robust model selection. To mitigate this problem, we propose a novel metric, the Unbiased Integration Coefficients (UIC), which exhibits significantly reduced bias ($p < 10^{-4}$) towards the minority class compared to conventional metrics. The UIC is constructed by aggregating existing metrics while penalising those more prone to imbalance. In addition, we introduce the Identical Partitions for Imbalance Problems (IPIP) algorithm for imbalanced ML problems, an ensemble-based approach. Our experimental results show that IPIP outperforms other baseline imbalance-aware approaches using Random Forest and Logistic Regression models in three out of seven datasets as assessed by the UIC metric, demonstrating its effectiveness in addressing imbalanced data challenges in binary classification tasks. This new framework for dealing with imbalanced datasets is materialized in the FILM (Framework for Imbalanced Learning Machines) R Package, accessible at https://github.com/antoniogt/FILM.
☆ Causally Reliable Concept Bottleneck Models
Concept-based models are an emerging paradigm in deep learning that constrains the inference process to operate through human-interpretable concepts, facilitating explainability and human interaction. However, these architectures, on par with popular opaque neural models, fail to account for the true causal mechanisms underlying the target phenomena represented in the data. This hampers their ability to support causal reasoning tasks, limits out-of-distribution generalization, and hinders the implementation of fairness constraints. To overcome these issues, we propose \emph{Causally reliable Concept Bottleneck Models} (C$^2$BMs), a class of concept-based architectures that enforce reasoning through a bottleneck of concepts structured according to a model of the real-world causal mechanisms. We also introduce a pipeline to automatically learn this structure from observational data and \emph{unstructured} background knowledge (e.g., scientific literature). Experimental evidence suggest that C$^2$BM are more interpretable, causally reliable, and improve responsiveness to interventions w.r.t. standard opaque and concept-based models, while maintaining their accuracy.
☆ A Generalist Cross-Domain Molecular Learning Framework for Structure-Based Drug Discovery
Structure-based drug discovery (SBDD) is a systematic scientific process that develops new drugs by leveraging the detailed physical structure of the target protein. Recent advancements in pre-trained models for biomolecules have demonstrated remarkable success across various biochemical applications, including drug discovery and protein engineering. However, in most approaches, the pre-trained models primarily focus on the characteristics of either small molecules or proteins, without delving into their binding interactions which are essential cross-domain relationships pivotal to SBDD. To fill this gap, we propose a general-purpose foundation model named BIT (an abbreviation for Biomolecular Interaction Transformer), which is capable of encoding a range of biochemical entities, including small molecules, proteins, and protein-ligand complexes, as well as various data formats, encompassing both 2D and 3D structures. Specifically, we introduce Mixture-of-Domain-Experts (MoDE) to handle the biomolecules from diverse biochemical domains and Mixture-of-Structure-Experts (MoSE) to capture positional dependencies in the molecular structures. The proposed mixture-of-experts approach enables BIT to achieve both deep fusion and domain-specific encoding, effectively capturing fine-grained molecular interactions within protein-ligand complexes. Then, we perform cross-domain pre-training on the shared Transformer backbone via several unified self-supervised denoising tasks. Experimental results on various benchmarks demonstrate that BIT achieves exceptional performance in downstream tasks, including binding affinity prediction, structure-based virtual screening, and molecular property prediction.
☆ Learning Causal Response Representations through Direct Effect Analysis
We propose a novel approach for learning causal response representations. Our method aims to extract directions in which a multidimensional outcome is most directly caused by a treatment variable. By bridging conditional independence testing with causal representation learning, we formulate an optimisation problem that maximises the evidence against conditional independence between the treatment and outcome, given a conditioning set. This formulation employs flexible regression models tailored to specific applications, creating a versatile framework. The problem is addressed through a generalised eigenvalue decomposition. We show that, under mild assumptions, the distribution of the largest eigenvalue can be bounded by a known $F$-distribution, enabling testable conditional independence. We also provide theoretical guarantees for the optimality of the learned representation in terms of signal-to-noise ratio and Fisher information maximisation. Finally, we demonstrate the empirical effectiveness of our approach in simulation and real-world experiments. Our results underscore the utility of this framework in uncovering direct causal effects within complex, multivariate settings.
comment: 32 pages, 15 figures, stat.ML
☆ scDD: Latent Codes Based scRNA-seq Dataset Distillation with Foundation Model Knowledge
Single-cell RNA sequencing (scRNA-seq) technology has profiled hundreds of millions of human cells across organs, diseases, development and perturbations to date. However, the high-dimensional sparsity, batch effect noise, category imbalance, and ever-increasing data scale of the original sequencing data pose significant challenges for multi-center knowledge transfer, data fusion, and cross-validation between scRNA-seq datasets. To address these barriers, (1) we first propose a latent codes-based scRNA-seq dataset distillation framework named scDD, which transfers and distills foundation model knowledge and original dataset information into a compact latent space and generates synthetic scRNA-seq dataset by a generator to replace the original dataset. Then, (2) we propose a single-step conditional diffusion generator named SCDG, which perform single-step gradient back-propagation to help scDD optimize distillation quality and avoid gradient decay caused by multi-step back-propagation. Meanwhile, SCDG ensures the scRNA-seq data characteristics and inter-class discriminability of the synthetic dataset through flexible conditional control and generation quality assurance. Finally, we propose a comprehensive benchmark to evaluate the performance of scRNA-seq dataset distillation in different data analysis tasks. It is validated that our proposed method can achieve 7.61% absolute and 15.70% relative improvement over previous state-of-the-art methods on average task.
☆ EDCA -- An Evolutionary Data-Centric AutoML Framework for Efficient Pipelines
Automated Machine Learning (AutoML) gained popularity due to the increased demand for Machine Learning (ML) specialists, allowing them to apply ML techniques effortlessly and quickly. AutoML implementations use optimisation methods to identify the most effective ML solution for a given dataset, aiming to improve one or more predefined metrics. However, most implementations focus on model selection and hyperparameter tuning. Despite being an important factor in obtaining high-performance ML systems, data quality is usually an overlooked part of AutoML and continues to be a manual and time-consuming task. This work presents EDCA, an Evolutionary Data Centric AutoML framework. In addition to the traditional tasks such as selecting the best models and hyperparameters, EDCA enhances the given data by optimising data processing tasks such as data reduction and cleaning according to the problems' needs. All these steps create an ML pipeline that is optimised by an evolutionary algorithm. To assess its effectiveness, EDCA was compared to FLAML and TPOT, two frameworks at the top of the AutoML benchmarks. The frameworks were evaluated in the same conditions using datasets from AMLB classification benchmarks. EDCA achieved statistically similar results in performance to FLAML and TPOT but used significantly less data to train the final solutions. Moreover, EDCA experimental results reveal that a good performance can be achieved using less data and efficient ML algorithm aspects that align with Green AutoML guidelines
Large Language Models for Zero-shot Inference of Causal Structures in Biology ICLR 2025
Genes, proteins and other biological entities influence one another via causal molecular networks. Causal relationships in such networks are mediated by complex and diverse mechanisms, through latent variables, and are often specific to cellular context. It remains challenging to characterise such networks in practice. Here, we present a novel framework to evaluate large language models (LLMs) for zero-shot inference of causal relationships in biology. In particular, we systematically evaluate causal claims obtained from an LLM using real-world interventional data. This is done over one hundred variables and thousands of causal hypotheses. Furthermore, we consider several prompting and retrieval-augmentation strategies, including large, and potentially conflicting, collections of scientific articles. Our results show that with tailored augmentation and prompting, even relatively small LLMs can capture meaningful aspects of causal structure in biological systems. This supports the notion that LLMs could act as orchestration tools in biological discovery, by helping to distil current knowledge in ways amenable to downstream analysis. Our approach to assessing LLMs with respect to experimental data is relevant for a broad range of problems at the intersection of causal learning, LLMs and scientific discovery.
comment: ICLR 2025 Workshop on Machine Learning for Genomics Explorations
☆ TRANSIT your events into a new mass: Fast background interpolation for weakly-supervised anomaly searches
We introduce a new model for conditional and continuous data morphing called TRansport Adversarial Network for Smooth InTerpolation (TRANSIT). We apply it to create a background data template for weakly-supervised searches at the LHC. The method smoothly transforms sideband events to match signal region mass distributions. We demonstrate the performance of TRANSIT using the LHC Olympics R\&D dataset. The model captures non-linear mass correlations of features and produces a template that offers a competitive anomaly sensitivity compared to state-of-the-art transport-based template generators. Moreover, the computational training time required for TRANSIT is an order of magnitude lower than that of competing deep learning methods. This makes it ideal for analyses that iterate over many signal regions and signal models. Unlike generative models, which must learn a full probability density distribution, i.e., the correlations between all the variables, the proposed transport model only has to learn a smooth conditional shift of the distribution. This allows for a simpler, more efficient residual architecture, enabling mass uncorrelated features to pass the network unchanged while the mass correlated features are adjusted accordingly. Furthermore, we show that the latent space of the model provides a set of mass decorrelated features useful for anomaly detection without background sculpting.
comment: 34 pages, 14 figures
☆ The Challenge of Identifying the Origin of Black-Box Large Language Models
The tremendous commercial potential of large language models (LLMs) has heightened concerns about their unauthorized use. Third parties can customize LLMs through fine-tuning and offer only black-box API access, effectively concealing unauthorized usage and complicating external auditing processes. This practice not only exacerbates unfair competition, but also violates licensing agreements. In response, identifying the origin of black-box LLMs is an intrinsic solution to this issue. In this paper, we first reveal the limitations of state-of-the-art passive and proactive identification methods with experiments on 30 LLMs and two real-world black-box APIs. Then, we propose the proactive technique, PlugAE, which optimizes adversarial token embeddings in a continuous space and proactively plugs them into the LLM for tracing and identification. The experiments show that PlugAE can achieve substantial improvement in identifying fine-tuned derivatives. We further advocate for legal frameworks and regulations to better address the challenges posed by the unauthorized use of LLMs.
☆ InFL-UX: A Toolkit for Web-Based Interactive Federated Learning
This paper presents InFL-UX, an interactive, proof-of-concept browser-based Federated Learning (FL) toolkit designed to integrate user contributions seamlessly into the machine learning (ML) workflow. InFL-UX enables users across multiple devices to upload datasets, define classes, and collaboratively train classification models directly in the browser using modern web technologies. Unlike traditional FL toolkits, which often focus on backend simulations, InFL-UX provides a simple user interface for researchers to explore how users interact with and contribute to FL systems in real-world, interactive settings. By prioritising usability and decentralised model training, InFL-UX bridges the gap between FL and Interactive Machine Learning (IML), empowering non-technical users to actively participate in ML classification tasks.
☆ Provable Robust Overfitting Mitigation in Wasserstein Distributionally Robust Optimization
Wasserstein distributionally robust optimization (WDRO) optimizes against worst-case distributional shifts within a specified uncertainty set, leading to enhanced generalization on unseen adversarial examples, compared to standard adversarial training which focuses on pointwise adversarial perturbations. However, WDRO still suffers fundamentally from the robust overfitting problem, as it does not consider statistical error. We address this gap by proposing a novel robust optimization framework under a new uncertainty set for adversarial noise via Wasserstein distance and statistical error via Kullback-Leibler divergence, called the Statistically Robust WDRO. We establish a robust generalization bound for the new optimization framework, implying that out-of-distribution adversarial performance is at least as good as the statistically robust training loss with high probability. Furthermore, we derive conditions under which Stackelberg and Nash equilibria exist between the learner and the adversary, giving an optimal robust model in certain sense. Finally, through extensive experiments, we demonstrate that our method significantly mitigates robust overfitting and enhances robustness within the framework of WDRO.
☆ Explainable AI in Time-Sensitive Scenarios: Prefetched Offline Explanation Model
As predictive machine learning models become increasingly adopted and advanced, their role has evolved from merely predicting outcomes to actively shaping them. This evolution has underscored the importance of Trustworthy AI, highlighting the necessity to extend our focus beyond mere accuracy and toward a comprehensive understanding of these models' behaviors within the specific contexts of their applications. To further progress in explainability, we introduce Poem, Prefetched Offline Explanation Model, a model-agnostic, local explainability algorithm for image data. The algorithm generates exemplars, counterexemplars and saliency maps to provide quick and effective explanations suitable for time-sensitive scenarios. Leveraging an existing local algorithm, \poem{} infers factual and counterfactual rules from data to create illustrative examples and opposite scenarios with an enhanced stability by design. A novel mechanism then matches incoming test points with an explanation base and produces diverse exemplars, informative saliency maps and believable counterexemplars. Experimental results indicate that Poem outperforms its predecessor Abele in speed and ability to generate more nuanced and varied exemplars alongside more insightful saliency maps and valuable counterexemplars.
☆ Towards Autonomous Reinforcement Learning for Real-World Robotic Manipulation with Large Language Models
Recent advancements in Large Language Models (LLMs) and Visual Language Models (VLMs) have significantly impacted robotics, enabling high-level semantic motion planning applications. Reinforcement Learning (RL), a complementary paradigm, enables agents to autonomously optimize complex behaviors through interaction and reward signals. However, designing effective reward functions for RL remains challenging, especially in real-world tasks where sparse rewards are insufficient and dense rewards require elaborate design. In this work, we propose Autonomous Reinforcement learning for Complex HumanInformed Environments (ARCHIE), an unsupervised pipeline leveraging GPT-4, a pre-trained LLM, to generate reward functions directly from natural language task descriptions. The rewards are used to train RL agents in simulated environments, where we formalize the reward generation process to enhance feasibility. Additionally, GPT-4 automates the coding of task success criteria, creating a fully automated, one-shot procedure for translating human-readable text into deployable robot skills. Our approach is validated through extensive simulated experiments on single-arm and bi-manual manipulation tasks using an ABB YuMi collaborative robot, highlighting its practicality and effectiveness. Tasks are demonstrated on the real robot setup.
☆ A General Framework for Scalable UE-AP Association in User-Centric Cell-Free Massive MIMO based on Recurrent Neural Networks
This study addresses the challenge of access point (AP) and user equipment (UE) association in cell-free massive MIMO networks. It introduces a deep learning algorithm leveraging Bidirectional Long Short-Term Memory cells and a hybrid probabilistic methodology for weight updating. This approach enhances scalability by adapting to variations in the number of UEs without requiring retraining. Additionally, the study presents a training methodology that improves scalability not only with respect to the number of UEs but also to the number of APs. Furthermore, a variant of the proposed AP-UE algorithm ensures robustness against pilot contamination effects, a critical issue arising from pilot reuse in channel estimation. Extensive numerical results validate the effectiveness and adaptability of the proposed methods, demonstrating their superiority over widely used heuristic alternatives.
comment: submitted to IEEE journal
☆ Frequency Hopping Synchronization by Reinforcement Learning for Satellite Communication System
Satellite communication systems (SCSs) used for tactical purposes require robust security and anti-jamming capabilities, making frequency hopping (FH) a powerful option. However, the current FH systems face challenges due to significant interference from other devices and the considerable path loss inherent in satellite communication. This misalignment leads to inefficient synchronization, crucial for maintaining reliable communication. Traditional methods, such as those employing long short-term memory (LSTM) networks, have made improvements, but they still struggle in dynamic conditions of satellite environments. This paper presents a novel method for synchronizing FH signals in tactical SCSs by combining serial search and reinforcement learning to achieve coarse and fine acquisition, respectively. The mathematical analysis and simulation results demonstrate that the proposed method reduces the average number of hops required for synchronization by 58.17% and mean squared error (MSE) of the uplink hop timing estimation by 76.95%, as compared to the conventional serial search method. Comparing with the early late gate synchronization method based on serial search and use of LSTM network, the average number of hops for synchronization is reduced by 12.24% and the MSE by 18.5%.
comment: 18pages, 5figures
☆ Bi-Lipschitz Ansatz for Anti-Symmetric Functions
Motivated by applications for simulating quantum many body functions, we propose a new universal ansatz for approximating anti-symmetric functions. The main advantage of this ansatz over previous alternatives is that it is bi-Lipschitz with respect to a naturally defined metric. As a result, we are able to obtain quantitative approximation results for approximation of Lipschitz continuous antisymmetric functions. Moreover, we provide preliminary experimental evidence to the improved performance of this ansatz for learning antisymmetric functions.
☆ Knowledge Retention for Continual Model-Based Reinforcement Learning
We propose DRAGO, a novel approach for continual model-based reinforcement learning aimed at improving the incremental development of world models across a sequence of tasks that differ in their reward functions but not the state space or dynamics. DRAGO comprises two key components: Synthetic Experience Rehearsal, which leverages generative models to create synthetic experiences from past tasks, allowing the agent to reinforce previously learned dynamics without storing data, and Regaining Memories Through Exploration, which introduces an intrinsic reward mechanism to guide the agent toward revisiting relevant states from prior tasks. Together, these components enable the agent to maintain a comprehensive and continually developing world model, facilitating more effective learning and adaptation across diverse environments. Empirical evaluations demonstrate that DRAGO is able to preserve knowledge across tasks, achieving superior performance in various continual learning scenarios.
☆ RCRank: Multimodal Ranking of Root Causes of Slow Queries in Cloud Database Systems VLDB 2025
With the continued migration of storage to cloud database systems,the impact of slow queries in such systems on services and user experience is increasing. Root-cause diagnosis plays an indispensable role in facilitating slow-query detection and revision. This paper proposes a method capable of both identifying possible root cause types for slow queries and ranking these according to their potential for accelerating slow queries. This enables prioritizing root causes with the highest impact, in turn improving slow-query revision effectiveness. To enable more accurate and detailed diagnoses, we propose the multimodal Ranking for the Root Causes of slow queries (RCRank) framework, which formulates root cause analysis as a multimodal machine learning problem and leverages multimodal information from query statements, execution plans, execution logs, and key performance indicators. To obtain expressive embeddings from its heterogeneous multimodal input, RCRank integrates self-supervised pre-training that enhances cross-modal alignment and task relevance. Next, the framework integrates root-cause-adaptive cross Transformers that enable adaptive fusion of multimodal features with varying characteristics. Finally, the framework offers a unified model that features an impact-aware training objective for identifying and ranking root causes. We report on experiments on real and synthetic datasets, finding that RCRank is capable of consistently outperforming the state-of-the-art methods at root cause identification and ranking according to a range of metrics.
comment: Accepted by VLDB 2025
☆ How to Mitigate Overfitting in Weak-to-strong Generalization?
Aligning powerful AI models on tasks that surpass human evaluation capabilities is the central problem of \textbf{superalignment}. To address this problem, weak-to-strong generalization aims to elicit the capabilities of strong models through weak supervisors and ensure that the behavior of strong models aligns with the intentions of weak supervisors without unsafe behaviors such as deception. Although weak-to-strong generalization exhibiting certain generalization capabilities, strong models exhibit significant overfitting in weak-to-strong generalization: Due to the strong fit ability of strong models, erroneous labels from weak supervisors may lead to overfitting in strong models. In addition, simply filtering out incorrect labels may lead to a degeneration in question quality, resulting in a weak generalization ability of strong models on hard questions. To mitigate overfitting in weak-to-strong generalization, we propose a two-stage framework that simultaneously improves the quality of supervision signals and the quality of input questions. Experimental results in three series of large language models and two mathematical benchmarks demonstrate that our framework significantly improves PGR compared to naive weak-to-strong generalization, even achieving up to 100\% PGR on some models.
☆ Incorporating Surrogate Gradient Norm to Improve Offline Optimization Techniques
Offline optimization has recently emerged as an increasingly popular approach to mitigate the prohibitively expensive cost of online experimentation. The key idea is to learn a surrogate of the black-box function that underlines the target experiment using a static (offline) dataset of its previous input-output queries. Such an approach is, however, fraught with an out-of-distribution issue where the learned surrogate becomes inaccurate outside the offline data regimes. To mitigate this, existing offline optimizers have proposed numerous conditioning techniques to prevent the learned surrogate from being too erratic. Nonetheless, such conditioning strategies are often specific to particular surrogate or search models, which might not generalize to a different model choice. This motivates us to develop a model-agnostic approach instead, which incorporates a notion of model sharpness into the training loss of the surrogate as a regularizer. Our approach is supported by a new theoretical analysis demonstrating that reducing surrogate sharpness on the offline dataset provably reduces its generalized sharpness on unseen data. Our analysis extends existing theories from bounding generalized prediction loss (on unseen data) with loss sharpness to bounding the worst-case generalized surrogate sharpness with its empirical estimate on training data, providing a new perspective on sharpness regularization. Our extensive experimentation on a diverse range of optimization tasks also shows that reducing surrogate sharpness often leads to significant improvement, marking (up to) a noticeable 9.6% performance boost. Our code is publicly available at https://github.com/cuong-dm/IGNITE
☆ ThrowBench: Benchmarking LLMs by Predicting Runtime Exceptions
Modern Large Language Models (LLMs) have shown astounding capabilities of code understanding and synthesis. In order to assess such capabilities, several benchmarks have been devised (e.g., HumanEval). However, most benchmarks focus on code synthesis from natural language instructions. Hence, such benchmarks do not test for other forms of code understanding. Moreover, there have been concerns about contamination and leakage. That is, benchmark problems (or closely related problems) may appear in training set, strongly biasing benchmark results. In this work we investigate whether large language models can correctly predict runtime program behavior. To this end, we introduce ThrowBench, a benchmark consisting of over 2,400 short user-written programs written in four different programming languages. The majority of these programs throw an exception during runtime (due to a bug). LLMs are asked to predict whether a presented program throws an exception and, if so, which one. Evaluating our benchmark on six state-of-the-art code LLMs we see modest performance ranging from 19 to 38% (F1 score). Benchmarking a wider set of code capabilities could improve the assessment of code LLMs and help identify weak points in current models. Moreover, as ground-truth answers have been determined through program execution, leakage is not a concern. We release ThrowBench as well as all of our results together with this work.
☆ One-Shot Clustering for Federated Learning
Federated Learning (FL) is a widespread and well adopted paradigm of decentralized learning that allows training one model from multiple sources without the need to directly transfer data between participating clients. Since its inception in 2015, it has been divided into numerous sub-fields that deal with application-specific issues, be it data heterogeneity or resource allocation. One such sub-field, Clustered Federated Learning (CFL), is dealing with the problem of clustering the population of clients into separate cohorts to deliver personalized models. Although few remarkable works have been published in this domain, the problem is still largely unexplored, as its basic assumption and settings are slightly different from standard FL. In this work, we present One-Shot Clustered Federated Learning (OCFL), a clustering-agnostic algorithm that can automatically detect the earliest suitable moment for clustering. Our algorithm is based on the computation of cosine similarity between gradients of the clients and a temperature measure that detects when the federated model starts to converge. We empirically evaluate our methodology by testing various one-shot clustering algorithms for over thirty different tasks on three benchmark datasets. Our experiments showcase the good performance of our approach when used to perform CFL in an automated manner without the need to adjust hyperparameters.
☆ Synthetic Data is an Elegant GIFT for Continual Vision-Language Models CVPR 2025
Pre-trained Vision-Language Models (VLMs) require Continual Learning (CL) to efficiently update their knowledge and adapt to various downstream tasks without retraining from scratch. However, for VLMs, in addition to the loss of knowledge previously learned from downstream tasks, pre-training knowledge is also corrupted during continual fine-tuning. This issue is exacerbated by the unavailability of original pre-training data, leaving VLM's generalization ability degrading. In this paper, we propose GIFT, a novel continual fine-tuning approach that utilizes synthetic data to overcome catastrophic forgetting in VLMs. Taking advantage of recent advances in text-to-image synthesis, we employ a pre-trained diffusion model to recreate both pre-training and learned downstream task data. In this way, the VLM can revisit previous knowledge through distillation on matching diffusion-generated images and corresponding text prompts. Leveraging the broad distribution and high alignment between synthetic image-text pairs in VLM's feature space, we propose a contrastive distillation loss along with an image-text alignment constraint. To further combat in-distribution overfitting and enhance distillation performance with limited amount of generated data, we incorporate adaptive weight consolidation, utilizing Fisher information from these synthetic image-text pairs and achieving a better stability-plasticity balance. Extensive experiments demonstrate that our method consistently outperforms previous state-of-the-art approaches across various settings.
comment: This work is accepted by CVPR 2025. Modifications may be performed
☆ Quantum-Inspired Reinforcement Learning in the Presence of Epistemic Ambivalence
The complexity of online decision-making under uncertainty stems from the requirement of finding a balance between exploiting known strategies and exploring new possibilities. Naturally, the uncertainty type plays a crucial role in developing decision-making strategies that manage complexity effectively. In this paper, we focus on a specific form of uncertainty known as epistemic ambivalence (EA), which emerges from conflicting pieces of evidence or contradictory experiences. It creates a delicate interplay between uncertainty and confidence, distinguishing it from epistemic uncertainty that typically diminishes with new information. Indeed, ambivalence can persist even after additional knowledge is acquired. To address this phenomenon, we propose a novel framework, called the epistemically ambivalent Markov decision process (EA-MDP), aiming to understand and control EA in decision-making processes. This framework incorporates the concept of a quantum state from the quantum mechanics formalism, and its core is to assess the probability and reward of every possible outcome. We calculate the reward function using quantum measurement techniques and prove the existence of an optimal policy and an optimal value function in the EA-MDP framework. We also propose the EA-epsilon-greedy Q-learning algorithm. To evaluate the impact of EA on decision-making and the expedience of our framework, we study two distinct experimental setups, namely the two-state problem and the lattice problem. Our results show that using our methods, the agent converges to the optimal policy in the presence of EA.
☆ FUSE: First-Order and Second-Order Unified SynthEsis in Stochastic Optimization
Stochastic optimization methods have actively been playing a critical role in modern machine learning algorithms to deliver decent performance. While numerous works have proposed and developed diverse approaches, first-order and second-order methods are in entirely different situations. The former is significantly pivotal and dominating in emerging deep learning but only leads convergence to a stationary point. However, second-order methods are less popular due to their computational intensity in large-dimensional problems. This paper presents a novel method that leverages both the first-order and second-order methods in a unified algorithmic framework, termed FUSE, from which a practical version (PV) is derived accordingly. FUSE-PV stands as a simple yet efficient optimization method involving a switch-over between first and second orders. Additionally, we develop different criteria that determine when to switch. FUSE-PV has provably shown a smaller computational complexity than SGD and Adam. To validate our proposed scheme, we present an ablation study on several simple test functions and show a comparison with baselines for benchmark datasets.
comment: 6 pages, 7 figures
☆ Geometric Re-Analysis of Classical MDP Solving Algorithms
We build on a recently introduced geometric interpretation of Markov Decision Processes (MDPs) to analyze classical MDP-solving algorithms: Value Iteration (VI) and Policy Iteration (PI). First, we develop a geometry-based analytical apparatus, including a transformation that modifies the discount factor $\gamma$, to improve convergence guarantees for these algorithms in several settings. In particular, one of our results identifies a rotation component in the VI method, and as a consequence shows that when a Markov Reward Process (MRP) induced by the optimal policy is irreducible and aperiodic, the asymptotic convergence rate of value iteration is strictly smaller than $\gamma$.
♻ ☆ Gated Delta Networks: Improving Mamba2 with Delta Rule ICLR 2025
Linear Transformers have gained attention as efficient alternatives to standard Transformers, but their performance in retrieval and long-context tasks has been limited. To address these limitations, recent work has explored two distinct mechanisms: gating for adaptive memory control and the delta update rule for precise memory modifications. We observe that these mechanisms are complementary: gating enables rapid memory erasure while the delta rule facilitates targeted updates. Building on this insight, we introduce the gated delta rule and develop a parallel training algorithm optimized for modern hardware. Our proposed architecture, Gated DeltaNet, consistently surpasses existing models like Mamba2 and DeltaNet across multiple benchmarks, including language modeling, common-sense reasoning, in-context retrieval, length extrapolation, and long-context understanding. We further enhance performance by developing hybrid architectures that combine Gated DeltaNet layers with sliding window attention or Mamba2 layers, achieving both improved training efficiency and superior task performance.
comment: ICLR 2025 camera ready
♻ ☆ When Can You Get Away with Low Memory Adam?
Adam is the go-to optimizer for training modern machine learning models, but it requires additional memory to maintain the moving averages of the gradients and their squares. While various low-memory optimizers have been proposed that sometimes match the performance of Adam, their lack of reliability has left Adam as the default choice. In this work, we apply a simple layer-wise Signal-to-Noise Ratio (SNR) analysis to quantify when second-moment tensors can be effectively replaced by their means across different dimensions. Our SNR analysis reveals how architecture, training hyperparameters, and dataset properties impact compressibility along Adam's trajectory, naturally leading to $\textit{SlimAdam}$, a memory-efficient Adam variant. $\textit{SlimAdam}$ compresses the second moments along dimensions with high SNR when feasible, and leaves when compression would be detrimental. Through experiments across a diverse set of architectures and training scenarios, we show that $\textit{SlimAdam}$ matches Adam's performance and stability while saving up to $98\%$ of total second moments. Code for $\textit{SlimAdam}$ is available at https://github.com/dayal-kalra/low-memory-adam.
comment: Acknowledgement updates and minor writing edits
♻ ☆ RAAD-LLM: Adaptive Anomaly Detection Using LLMs and RAG Integration
Anomaly detection in complex industrial environments poses unique challenges, particularly in contexts characterized by data sparsity and evolving operational conditions. Predictive maintenance (PdM) in such settings demands methodologies that are adaptive, transferable, and capable of integrating domain-specific knowledge. In this paper, we present RAAD-LLM, a novel framework for adaptive anomaly detection, leveraging large language models (LLMs) integrated with Retrieval-Augmented Generation (RAG). This approach addresses the aforementioned PdM challenges. By effectively utilizing domain-specific knowledge, RAAD-LLM enhances the detection of anomalies in time series data without requiring fine-tuning on specific datasets. The framework's adaptability mechanism enables it to adjust its understanding of normal operating conditions dynamically, thus increasing detection accuracy. We validate this methodology through a real-world application for a plastics manufacturing plant and the Skoltech Anomaly Benchmark (SKAB). Results show significant improvements over our previous model with an accuracy increase from 70.7% to 89.1% on the real-world dataset. By allowing for the enriching of input series data with semantics, RAAD-LLM incorporates multimodal capabilities that facilitate more collaborative decision-making between the model and plant operators. Overall, our findings support RAAD-LLM's ability to revolutionize anomaly detection methodologies in PdM, potentially leading to a paradigm shift in how anomaly detection is implemented across various industries.
comment: arXiv admin note: substantial text overlap with arXiv:2411.00914
♻ ☆ The Last Iterate Advantage: Empirical Auditing and Principled Heuristic Analysis of Differentially Private SGD ICLR 2025
We propose a simple heuristic privacy analysis of noisy clipped stochastic gradient descent (DP-SGD) in the setting where only the last iterate is released and the intermediate iterates remain hidden. Namely, our heuristic assumes a linear structure for the model. We show experimentally that our heuristic is predictive of the outcome of privacy auditing applied to various training procedures. Thus it can be used prior to training as a rough estimate of the final privacy leakage. We also probe the limitations of our heuristic by providing some artificial counterexamples where it underestimates the privacy leakage. The standard composition-based privacy analysis of DP-SGD effectively assumes that the adversary has access to all intermediate iterates, which is often unrealistic. However, this analysis remains the state of the art in practice. While our heuristic does not replace a rigorous privacy analysis, it illustrates the large gap between the best theoretical upper bounds and the privacy auditing lower bounds and sets a target for further work to improve the theoretical privacy analyses. We also empirically support our heuristic and show existing privacy auditing attacks are bounded by our heuristic analysis in both vision and language tasks.
comment: ICLR 2025 camera-ready version
♻ ☆ Some Targets Are Harder to Identify than Others: Quantifying the Target-dependent Membership Leakage AISTATS 2025
In a Membership Inference (MI) game, an attacker tries to infer whether a target point was included or not in the input of an algorithm. Existing works show that some target points are easier to identify, while others are harder. This paper explains the target-dependent hardness of membership attacks by studying the powers of the optimal attacks in a fixed-target MI game. We characterise the optimal advantage and trade-off functions of attacks against the empirical mean in terms of the Mahalanobis distance between the target point and the data-generating distribution. We further derive the impacts of two privacy defences, i.e. adding Gaussian noise and sub-sampling, and that of target misspecification on optimal attacks. As by-products of our novel analysis of the Likelihood Ratio (LR) test, we provide a new covariance attack which generalises and improves the scalar product attack. Also, we propose a new optimal canary-choosing strategy for auditing privacy in the white-box federated learning setting. Our experiments validate that the Mahalanobis score explains the hardness of fixed-target MI games.
comment: Appears in AISTATS 2025 (Oral)
♻ ☆ AdaptBot: Combining LLM with Knowledge Graphs and Human Input for Generic-to-Specific Task Decomposition and Knowledge Refinement ICRA
An embodied agent assisting humans is often asked to complete new tasks, and there may not be sufficient time or labeled examples to train the agent to perform these new tasks. Large Language Models (LLMs) trained on considerable knowledge across many domains can be used to predict a sequence of abstract actions for completing such tasks, although the agent may not be able to execute this sequence due to task-, agent-, or domain-specific constraints. Our framework addresses these challenges by leveraging the generic predictions provided by LLM and the prior domain knowledge encoded in a Knowledge Graph (KG), enabling an agent to quickly adapt to new tasks. The robot also solicits and uses human input as needed to refine its existing knowledge. Based on experimental evaluation in the context of cooking and cleaning tasks in simulation domains, we demonstrate that the interplay between LLM, KG, and human input leads to substantial performance gains compared with just using the LLM. Project website{\S}: https://sssshivvvv.github.io/adaptbot/
comment: Accepted to IEEE International Conference on Robotics and Automation (ICRA) 2025
♻ ☆ Detecting Systematic Weaknesses in Vision Models along Predefined Human-Understandable Dimensions
Slice discovery methods (SDMs) are prominent algorithms for finding systematic weaknesses in DNNs. They identify top-k semantically coherent slices/subsets of data where a DNN-under-test has low performance. For being directly useful, slices should be aligned with human-understandable and relevant dimensions, which, for example, are defined by safety and domain experts as part of the operational design domain (ODD). While SDMs can be applied effectively on structured data, their application on image data is complicated by the lack of semantic metadata. To address these issues, we present an algorithm that combines foundation models for zero-shot image classification to generate semantic metadata with methods for combinatorial search to find systematic weaknesses in images. In contrast to existing approaches, ours identifies weak slices that are in line with pre-defined human-understandable dimensions. As the algorithm includes foundation models, its intermediate and final results may not always be exact. Therefore, we include an approach to address the impact of noisy metadata. We validate our algorithm on both synthetic and real-world datasets, demonstrating its ability to recover human-understandable systematic weaknesses. Furthermore, using our approach, we identify systematic weaknesses of multiple pre-trained and publicly available state-of-the-art computer vision DNNs.
♻ ☆ Back Home: A Machine Learning Approach to Seashell Classification and Ecosystem Restoration
In Costa Rica, an average of 5 tons of seashells are extracted from ecosystems annually. Confiscated seashells, cannot be returned to their ecosystems due to the lack of origin recognition. To address this issue, we developed a convolutional neural network (CNN) specifically for seashell identification. We built a dataset from scratch, consisting of approximately 19000 images from the Pacific and Caribbean coasts. Using this dataset, the model achieved a classification accuracy exceeding 85%. The model has been integrated into a user-friendly application, which has classified over 36,000 seashells to date, delivering real-time results within 3 seconds per image. To further enhance the system's accuracy, an anomaly detection mechanism was incorporated to filter out irrelevant or anomalous inputs, ensuring only valid seashell images are processed.
♻ ☆ Tutorial on amortized optimization
Optimization is a ubiquitous modeling tool and is often deployed in settings which repeatedly solve similar instances of the same problem. Amortized optimization methods use learning to predict the solutions to problems in these settings, exploiting the shared structure between similar problem instances. These methods have been crucial in variational inference and reinforcement learning and are capable of solving optimization problems many orders of magnitudes times faster than traditional optimization methods that do not use amortization. This tutorial presents an introduction to the amortized optimization foundations behind these advancements and overviews their applications in variational inference, sparse coding, gradient-based meta-learning, control, reinforcement learning, convex optimization, optimal transport, and deep equilibrium networks. The source code for this tutorial is available at https://github.com/facebookresearch/amortized-optimization-tutorial.
comment: Foundations and Trends in Machine Learning
♻ ☆ A Simple and Effective Reinforcement Learning Method for Text-to-Image Diffusion Fine-tuning
Reinforcement learning (RL)-based fine-tuning has emerged as a powerful approach for aligning diffusion models with black-box objectives. Proximal policy optimization (PPO) is the most popular choice of method for policy optimization. While effective in terms of performance, PPO is highly sensitive to hyper-parameters and involves substantial computational overhead. REINFORCE, on the other hand, mitigates some computational complexities such as high memory overhead and sensitive hyper-parameter tuning, but has suboptimal performance due to high-variance and sample inefficiency. While the variance of the REINFORCE can be reduced by sampling multiple actions per input prompt and using a baseline correction term, it still suffers from sample inefficiency. To address these challenges, we systematically analyze the efficiency-effectiveness trade-off between REINFORCE and PPO, and propose leave-one-out PPO (LOOP), a novel RL for diffusion fine-tuning method. LOOP combines variance reduction techniques from REINFORCE, such as sampling multiple actions per input prompt and a baseline correction term, with the robustness and sample efficiency of PPO via clipping and importance sampling. Our results demonstrate that LOOP effectively improves diffusion models on various black-box objectives, and achieves a better balance between computational efficiency and performance.
♻ ☆ Human-Feedback Efficient Reinforcement Learning for Online Diffusion Model Finetuning ICLR
Controllable generation through Stable Diffusion (SD) fine-tuning aims to improve fidelity, safety, and alignment with human guidance. Existing reinforcement learning from human feedback methods usually rely on predefined heuristic reward functions or pretrained reward models built on large-scale datasets, limiting their applicability to scenarios where collecting such data is costly or difficult. To effectively and efficiently utilize human feedback, we develop a framework, HERO, which leverages online human feedback collected on the fly during model learning. Specifically, HERO features two key mechanisms: (1) Feedback-Aligned Representation Learning, an online training method that captures human feedback and provides informative learning signals for fine-tuning, and (2) Feedback-Guided Image Generation, which involves generating images from SD's refined initialization samples, enabling faster convergence towards the evaluator's intent. We demonstrate that HERO is 4x more efficient in online feedback for body part anomaly correction compared to the best existing method. Additionally, experiments show that HERO can effectively handle tasks like reasoning, counting, personalization, and reducing NSFW content with only 0.5K online feedback.
comment: Published in International Conference on Learning Representations (ICLR) 2025
♻ ☆ A learning-based approach to stochastic optimal control under reach-avoid constraint
We develop a model-free approach to optimally control stochastic, Markovian systems subject to a reach-avoid constraint. Specifically, the state trajectory must remain within a safe set while reaching a target set within a finite time horizon. Due to the time-dependent nature of these constraints, we show that, in general, the optimal policy for this constrained stochastic control problem is non-Markovian, which increases the computational complexity. To address this challenge, we apply the state-augmentation technique from arXiv:2402.19360, reformulating the problem as a constrained Markov decision process (CMDP) on an extended state space. This transformation allows us to search for a Markovian policy, avoiding the complexity of non-Markovian policies. To learn the optimal policy without a system model, and using only trajectory data, we develop a log-barrier policy gradient approach. We prove that under suitable assumptions, the policy parameters converge to the optimal parameters, while ensuring that the system trajectories satisfy the stochastic reach-avoid constraint with high probability.
♻ ☆ Towards One Model for Classical Dimensionality Reduction: A Probabilistic Perspective on UMAP and t-SNE
This paper shows that dimensionality reduction methods such as UMAP and t-SNE, can be approximately recast as MAP inference methods corresponding to a model introduced in ProbDR, that describes the graph Laplacian (an estimate of the data precision matrix) using a Wishart distribution, with a mean given by a non-linear covariance function evaluated on the latents. This interpretation offers deeper theoretical and semantic insights into such algorithms, by showing that variances corresponding to these covariances are low (potentially misspecified), and forging a connection to Gaussian process latent variable models by showing that well-known kernels can be used to describe covariances implied by graph Laplacians. We also introduce tools with which similar dimensionality reduction methods can be studied.
comment: Updated preprint
♻ ☆ Protein Large Language Models: A Comprehensive Survey
Protein-specific large language models (Protein LLMs) are revolutionizing protein science by enabling more efficient protein structure prediction, function annotation, and design. While existing surveys focus on specific aspects or applications, this work provides the first comprehensive overview of Protein LLMs, covering their architectures, training datasets, evaluation metrics, and diverse applications. Through a systematic analysis of over 100 articles, we propose a structured taxonomy of state-of-the-art Protein LLMs, analyze how they leverage large-scale protein sequence data for improved accuracy, and explore their potential in advancing protein engineering and biomedical research. Additionally, we discuss key challenges and future directions, positioning Protein LLMs as essential tools for scientific discovery in protein science. Resources are maintained at https://github.com/Yijia-Xiao/Protein-LLM-Survey.
comment: 24 pages, 4 figures, 5 tables
♻ ☆ Streaming Private Continual Counting via Binning
In differential privacy, $\textit{continual observation}$ refers to problems in which we wish to continuously release a function of a dataset that is revealed one element at a time. The challenge is to maintain a good approximation while keeping the combined output over all time steps differentially private. In the special case of $\textit{continual counting}$ we seek to approximate a sum of binary input elements. This problem has received considerable attention lately, in part due to its relevance in implementations of differentially private stochastic gradient descent. $\textit{Factorization mechanisms}$ are the leading approach to continual counting, but the best such mechanisms do not work well in $\textit{streaming}$ settings since they require space proportional to the size of the input. In this paper, we present a simple approach to approximating factorization mechanisms in low space via $\textit{binning}$, where adjacent matrix entries with similar values are changed to be identical in such a way that a matrix-vector product can be maintained in sublinear space. Our approach has provable sublinear space guarantees for a class of lower triangular matrices whose entries are monotonically decreasing away from the diagonal. We show empirically that even with very low space usage we are able to closely match, and sometimes surpass, the performance of asymptotically optimal factorization mechanisms. Recently, and independently of our work, Dvijotham et al. have also suggested an approach to implementing factorization mechanisms in a streaming setting. Their work differs from ours in several respects: It only addresses factorization into $\textit{Toeplitz}$ matrices, only considers $\textit{maximum}$ error, and uses a different technique based on rational function approximation that seems less versatile than our binning approach.
comment: Accepted to SaTML 2025. Final version to appear on IEEE eXplore
♻ ☆ $\texttt{SEM-CTRL}$: Semantically Controlled Decoding
Ensuring both syntactic and semantic correctness in Large Language Model (LLM) outputs remains a significant challenge, despite being critical for real-world deployment. In this paper, we introduce $\texttt{SEM-CTRL}$, a unified approach that enforces rich context-sensitive constraints and task- and instance-specific semantics directly on an LLM decoder. Our approach integrates token-level MCTS, which is guided by specific syntactic and semantic constraints. The constraints over the desired outputs are expressed using Answer Set Grammars -- a logic-based formalism that generalizes context-sensitive grammars while incorporating background knowledge to represent task-specific semantics. We show that our approach guarantees correct completions for any off-the-shelf LLM without the need for fine-tuning. We evaluate $\texttt{SEM-CTRL}$ on a range of tasks, including synthetic grammar synthesis, combinatorial reasoning, and planning. Our results demonstrate that $\texttt{SEM-CTRL}$ allows small pre-trained LLMs to efficiently outperform larger variants and state-of-the-art reasoning models (e.g., o1-preview) while simultaneously guaranteeing solution correctness.
♻ ☆ Golden Ratio Weighting Prevents Model Collapse
Recent studies identified an intriguing phenomenon in recursive generative model training known as model collapse, where models trained on data generated by previous models exhibit severe performance degradation. Addressing this issue and developing more effective training strategies have become central challenges in generative model research. In this paper, we investigate this phenomenon theoretically within a novel framework, where generative models are iteratively trained on a combination of newly collected real data and synthetic data from the previous training step. To develop an optimal training strategy for integrating real and synthetic data, we evaluate the performance of a weighted training scheme in various scenarios, including Gaussian distribution estimation and linear regression. We theoretically characterize the impact of the mixing proportion and weighting scheme of synthetic data on the final model's performance. Our key finding is that, across different settings, the optimal weighting scheme under different proportions of synthetic data asymptotically follows a unified expression, revealing a fundamental trade-off between leveraging synthetic data and generative model performance. Notably, in some cases, the optimal weight assigned to real data corresponds to the reciprocal of the golden ratio. Finally, we validate our theoretical results on extensive simulated datasets and a real tabular dataset.
♻ ☆ Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution ICLR 2025
Probing learned concepts in large language models (LLMs) is crucial for understanding how semantic knowledge is encoded internally. Training linear classifiers on probing tasks is a principle approach to denote the vector of a certain concept in the representation space. However, the single vector identified for a concept varies with both data and training, making it less robust and weakening its effectiveness in real-world applications. To address this challenge, we propose an approach to approximate the subspace representing a specific concept. Built on linear probing classifiers, we extend the concept vectors into Gaussian Concept Subspace (GCS). We demonstrate GCS's effectiveness through measuring its faithfulness and plausibility across multiple LLMs with different sizes and architectures. Additionally, we use representation intervention tasks to showcase its efficacy in real-world applications such as emotion steering. Experimental results indicate that GCS concept vectors have the potential to balance steering performance and maintaining the fluency in natural language generation tasks.
comment: Accepted by ICLR 2025
♻ ☆ An Analysis Framework for Understanding Deep Neural Networks Based on Network Dynamics
Advancing artificial intelligence demands a deeper understanding of the mechanisms underlying deep learning. Here, we propose a straightforward analysis framework based on the dynamics of learning models. Neurons are categorized into two modes based on whether their transformation functions preserve order. This categorization reveals how deep neural networks (DNNs) maximize information extraction by rationally allocating the proportion of neurons in different modes across deep layers. We further introduce the attraction basins of the training samples in both the sample vector space and the weight vector space to characterize the generalization ability of DNNs. This framework allows us to identify optimal depth and width configurations, providing a unified explanation for fundamental DNN behaviors such as the "flat minima effect," "grokking," and double descent phenomena. Our analysis extends to networks with depths up to 100 layers.
comment: 12 pages, 10 figures
♻ ☆ Graph Neural Networks for Virtual Sensing in Complex Systems: Addressing Heterogeneous Temporal Dynamics SP
Real-time condition monitoring is crucial for the reliable and efficient operation of complex systems. However, relying solely on physical sensors can be limited due to their cost, placement constraints, or inability to directly measure certain critical parameters. Virtual sensing addresses these limitations by leveraging readily available sensor data and system knowledge to estimate inaccessible parameters or infer system states. The increasing complexity of industrial systems necessitates deployments of sensors with diverse modalities to provide a comprehensive understanding of system states. These sensors capture data at varying frequencies to monitor both rapid and slowly varying system dynamics, as well as local and global state evolutions of the systems. This leads to heterogeneous temporal dynamics, which, particularly under varying operational end environmental conditions, pose a significant challenge for accurate virtual sensing. To address this, we propose a Heterogeneous Temporal Graph Neural Network (HTGNN) framework. HTGNN explicitly models signals from diverse sensors and integrates operating conditions into the model architecture. We evaluate HTGNN using two newly released datasets: a bearing dataset with diverse load conditions for bearing load prediction and a year-long simulated dataset for predicting bridge live loads. Our results demonstrate that HTGNN significantly outperforms established baseline methods in both tasks, particularly under highly varying operating conditions. These results highlight HTGNN's potential as a robust and accurate virtual sensing approach for complex systems, paving the way for improved monitoring, predictive maintenance, and enhanced system performance. Our code and data are available under https://github.com/EPFL-IMOS/htgnn.
comment: This paper extends our previous conference paper (Best Paper at European Conference of the PHM Society 2024, https://doi.org/10.36001/phme.2024.v8i1.3998). Accepted by Mechanical Systems and Signal Processing (MSSP)
♻ ☆ The FFT Strikes Back: An Efficient Alternative to Self-Attention
Conventional self-attention mechanisms incur quadratic complexity, limiting their scalability on long sequences. We introduce \textbf{FFTNet}, an adaptive spectral filtering framework that leverages the Fast Fourier Transform (FFT) to achieve global token mixing in $\mathcal{O}(n\log n)$ time. By transforming inputs into the frequency domain, FFTNet exploits the orthogonality and energy preservation guaranteed by Parseval's theorem to capture long-range dependencies efficiently. Our main theoretical contributions are 1) an adaptive spectral filter, 2) combining local windowing with a global FFT branch, and 3) rich nonlinearity introduction in both the frequency and token domains. Experiments on the Long Range Arena and ImageNet benchmarks validate our theoretical insights and demonstrate superior performance over fixed Fourier and standard attention models.
♻ ☆ X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability
Despite the rapid development of safety alignment techniques for LLMs, defending against multi-turn jailbreaks is still a challenging task. In this paper, we conduct a comprehensive comparison, revealing that some existing defense methods can improve the robustness of LLMs against multi-turn jailbreaks but compromise usability, i.e., reducing general capabilities or causing the over-refusal problem. From the perspective of mechanism interpretability of LLMs, we discover that these methods fail to establish a boundary that exactly distinguishes safe and harmful feature representations. Therefore, boundary-safe representations close to harmful representations are inevitably disrupted, leading to a decline in usability. To address this issue, we propose X-Boundary to push harmful representations away from boundary-safe representations and obtain an exact distinction boundary. In this way, harmful representations can be precisely erased without disrupting safe ones. Experimental results show that X-Boundary achieves state-of-the-art defense performance against multi-turn jailbreaks, while reducing the over-refusal rate by about 20% and maintaining nearly complete general capability. Furthermore, we theoretically prove and empirically verify that X-Boundary can accelerate the convergence process during training. Please see our code at: https://github.com/AI45Lab/X-Boundary.
♻ ☆ Chunking the Critic: A Transformer-based Soft Actor-Critic with N-Step Returns
Soft Actor-Critic (SAC) critically depends on its critic network, which typically evaluates a single state-action pair to guide policy updates. Using N-step returns is a common practice to reduce the bias in the target values of the critic. However, using N-step returns can again introduce high variance and necessitates importance sampling, often destabilizing training. Recent algorithms have also explored action chunking-such as direct action repetition and movement primitives-to enhance exploration. In this paper, we propose a Transformer-based Critic Network for SAC that integrates the N-returns framework in a stable and efficient manner. Unlike approaches that perform chunking in the actor network, we feed chunked actions into the critic network to explore potential performance gains. Our architecture leverages the Transformer's ability to process sequential information, facilitating more robust value estimation. Empirical results show that this method not only achieves efficient, stable training but also excels in sparse reward/multi-phase environments-traditionally a challenge for step-based methods. These findings underscore the promise of combining Transformer-based critics with N-returns to advance reinforcement learning performance
♻ ☆ Learning finitely correlated states: stability of the spectral reconstruction
Matrix product operators allow efficient descriptions (or realizations) of states on a 1D lattice. We consider the task of learning a realization of minimal dimension from copies of an unknown state, such that the resulting operator is close to the density matrix in trace norm. For finitely correlated translation-invariant states on an infinite chain, a realization of minimal dimension can be exactly reconstructed via linear algebra operations from the marginals of a size depending on the representation dimension. We establish a bound on the trace norm error for an algorithm that estimates a candidate realization from estimates of these marginals and outputs a matrix product operator, estimating the state of a chain of arbitrary length $t$. This bound allows us to establish an $O(t^2)$ upper bound on the sample complexity of the learning task, with an explicit dependence on the site dimension, realization dimension and spectral properties of a certain map constructed from the state. A refined error bound can be proven for $C^*$-finitely correlated states, which have an operational interpretation in terms of sequential quantum channels applied to the memory system. We can also obtain an analogous error bound for a class of matrix product density operators on a finite chain reconstructible by local marginals. In this case, a linear number of marginals must be estimated, obtaining a sample complexity of $\tilde{O}(t^3)$. The learning algorithm also works for states that are sufficiently close to a finitely correlated state, with the potential of providing competitive algorithms for other interesting families of states.
comment: 42 pages, 7 figures. Manuscript restructured, with minor corrections and clarifications
♻ ☆ On the Challenges and Opportunities in Generative AI
The field of deep generative modeling has grown rapidly in the last few years. With the availability of massive amounts of training data coupled with advances in scalable unsupervised learning paradigms, recent large-scale generative models show tremendous promise in synthesizing high-resolution images and text, as well as structured data such as videos and molecules. However, we argue that current large-scale generative AI models exhibit several fundamental shortcomings that hinder their widespread adoption across domains. In this work, our objective is to identify these issues and highlight key unresolved challenges in modern generative AI paradigms that should be addressed to further enhance their capabilities, versatility, and reliability. By identifying these challenges, we aim to provide researchers with insights for exploring fruitful research directions, thus fostering the development of more robust and accessible generative AI solutions.
♻ ☆ Gumbel Counterfactual Generation From Language Models ICLR 2025
Understanding and manipulating the causal generation mechanisms in language models is essential for controlling their behavior. Previous work has primarily relied on techniques such as representation surgery -- e.g., model ablations or manipulation of linear subspaces tied to specific concepts -- to \emph{intervene} on these models. To understand the impact of interventions precisely, it is useful to examine \emph{counterfactuals} -- e.g., how a given sentence would have appeared had it been generated by the model following a specific intervention. We highlight that counterfactual reasoning is conceptually distinct from interventions, as articulated in Pearl's causal hierarchy. Based on this observation, we propose a framework for generating true string counterfactuals by reformulating language models as a structural equation model using the Gumbel-max trick, which we called Gumbel counterfactual generation. This reformulation allows us to model the joint distribution over original strings and their counterfactuals resulting from the same instantiation of the sampling noise. We develop an algorithm based on hindsight Gumbel sampling that allows us to infer the latent noise variables and generate counterfactuals of observed strings. Our experiments demonstrate that the approach produces meaningful counterfactuals while at the same time showing that commonly used intervention techniques have considerable undesired side effects.
comment: Accepted in ICLR 2025
Efficient Diversity-Preserving Diffusion Alignment via Gradient-Informed GFlowNets ICLR 2025
While one commonly trains large diffusion models by collecting datasets on target downstream tasks, it is often desired to align and finetune pretrained diffusion models with some reward functions that are either designed by experts or learned from small-scale datasets. Existing post-training methods for reward finetuning of diffusion models typically suffer from lack of diversity in generated samples, lack of prior preservation, and/or slow convergence in finetuning. Inspired by recent successes in generative flow networks (GFlowNets), a class of probabilistic models that sample with the unnormalized density of a reward function, we propose a novel GFlowNet method dubbed Nabla-GFlowNet (abbreviated as \methodname), the first GFlowNet method that leverages the rich signal in reward gradients, together with an objective called \graddb plus its variant \resgraddb designed for prior-preserving diffusion finetuning. We show that our proposed method achieves fast yet diversity- and prior-preserving finetuning of Stable Diffusion, a large-scale text-conditioned image diffusion model, on different realistic reward functions.
comment: Technical Report (35 pages, 31 figures), Accepted at ICLR 2025
♻ ☆ Procedural Knowledge in Pretraining Drives Reasoning in Large Language Models ICLR 2025
The capabilities and limitations of Large Language Models have been sketched out in great detail in recent years, providing an intriguing yet conflicting picture. On the one hand, LLMs demonstrate a general ability to solve problems. On the other hand, they show surprising reasoning gaps when compared to humans, casting doubt on the robustness of their generalisation strategies. The sheer volume of data used in the design of LLMs has precluded us from applying the method traditionally used to measure generalisation: train-test set separation. To overcome this, we study what kind of generalisation strategies LLMs employ when performing reasoning tasks by investigating the pretraining data they rely on. For two models of different sizes (7B and 35B) and 2.5B of their pretraining tokens, we identify what documents influence the model outputs for three simple mathematical reasoning tasks and contrast this to the data that are influential for answering factual questions. We find that, while the models rely on mostly distinct sets of data for each factual question, a document often has a similar influence across different reasoning questions within the same task, indicating the presence of procedural knowledge. We further find that the answers to factual questions often show up in the most influential data. However, for reasoning questions the answers usually do not show up as highly influential, nor do the answers to the intermediate reasoning steps. When we characterise the top ranked documents for the reasoning questions qualitatively, we confirm that the influential documents often contain procedural knowledge, like demonstrating how to obtain a solution using formulae or code. Our findings indicate that the approach to reasoning the models use is unlike retrieval, and more like a generalisable strategy that synthesises procedural knowledge from documents doing a similar form of reasoning.
comment: Published at ICLR 2025
♻ ☆ Learning from negative feedback, or positive feedback or both
Existing preference optimization methods often assume scenarios where paired preference feedback (preferred/positive vs. dis-preferred/negative examples) is available. This requirement limits their applicability in scenarios where only unpaired feedback--for example, either positive or negative--is available. To address this, we introduce a novel approach that decouples learning from positive and negative feedback. This decoupling enables control over the influence of each feedback type and, importantly, allows learning even when only one feedback type is present. A key contribution is demonstrating stable learning from negative feedback alone, a capability not well-addressed by current methods. Our approach builds upon the probabilistic framework introduced in (Dayan and Hinton, 1997), which uses expectation-maximization (EM) to directly optimize the probability of positive outcomes (as opposed to classic expected reward maximization). We address a key limitation in current EM-based methods: they solely maximize the likelihood of positive examples, while neglecting negative ones. We show how to extend EM algorithms to explicitly incorporate negative examples, leading to a theoretically grounded algorithm that offers an intuitive and versatile way to learn from both positive and negative feedback. We evaluate our approach for training language models based on human feedback as well as training policies for sequential decision-making problems, where learned value functions are available.
♻ ☆ MMGDreamer: Mixed-Modality Graph for Geometry-Controllable 3D Indoor Scene Generation AAAI 2025
Controllable 3D scene generation has extensive applications in virtual reality and interior design, where the generated scenes should exhibit high levels of realism and controllability in terms of geometry. Scene graphs provide a suitable data representation that facilitates these applications. However, current graph-based methods for scene generation are constrained to text-based inputs and exhibit insufficient adaptability to flexible user inputs, hindering the ability to precisely control object geometry. To address this issue, we propose MMGDreamer, a dual-branch diffusion model for scene generation that incorporates a novel Mixed-Modality Graph, visual enhancement module, and relation predictor. The mixed-modality graph allows object nodes to integrate textual and visual modalities, with optional relationships between nodes. It enhances adaptability to flexible user inputs and enables meticulous control over the geometry of objects in the generated scenes. The visual enhancement module enriches the visual fidelity of text-only nodes by constructing visual representations using text embeddings. Furthermore, our relation predictor leverages node representations to infer absent relationships between nodes, resulting in more coherent scene layouts. Extensive experimental results demonstrate that MMGDreamer exhibits superior control of object geometry, achieving state-of-the-art scene generation performance. Project page: https://yangzhifeio.github.io/project/MMGDreamer.
comment: Accepted by AAAI 2025 Main Track
♻ ☆ Magnetic Field Data Calibration with Transformer Model Using Physical Constraints: A Scalable Method for Satellite Missions, Illustrated by Tianwen-1
This study introduces a novel approach that integrates the magnetic field data correction from the Tianwen-1 Mars mission with a neural network architecture constrained by physical principles derived from Maxwell's equation equations. By employing a Transformer based model capable of efficiently handling sequential data, the method corrects measurement anomalies caused by satellite dynamics, instrument interference, and environmental noise. As a result, it significantly improves both the accuracy and the physical consistency of the calibrated data. Compared to traditional methods that require long data segments and manual intervention often taking weeks or even months to complete this new approach can finish calibration in just minutes to hours, and predictions are made within seconds. This innovation not only accelerates the process of space weather modeling and planetary magnetospheric studies but also provides a robust framework for future planetary exploration and solar wind interaction research.
♻ ☆ MobileViM: A Light-weight and Dimension-independent Vision Mamba for 3D Medical Image Analysis
Efficient evaluation of three-dimensional (3D) medical images is crucial for diagnostic and therapeutic practices in healthcare. Recent years have seen a substantial uptake in applying deep learning and computer vision to analyse and interpret medical images. Traditional approaches, such as convolutional neural networks (CNNs) and vision transformers (ViTs), face significant computational challenges, prompting the need for architectural advancements. Recent efforts have led to the introduction of novel architectures like the ``Mamba'' model as alternative solutions to traditional CNNs or ViTs. The Mamba model excels in the linear processing of one-dimensional data with low computational demands. However, Mamba's potential for 3D medical image analysis remains underexplored and could face significant computational challenges as the dimension increases. This manuscript presents MobileViM, a streamlined architecture for efficient segmentation of 3D medical images. In the MobileViM network, we invent a new dimension-independent mechanism and a dual-direction traversing approach to incorporate with a vision-Mamba-based framework. MobileViM also features a cross-scale bridging technique to improve efficiency and accuracy across various medical imaging modalities. With these enhancements, MobileViM achieves segmentation speeds exceeding 90 frames per second (FPS) on a single graphics processing unit (i.e., NVIDIA RTX 4090). This performance is over 24 FPS faster than the state-of-the-art deep learning models for processing 3D images with the same computational resources. In addition, experimental evaluations demonstrate that MobileViM delivers superior performance, with Dice similarity scores reaching 92.72%, 86.69%, 80.46%, and 77.43% for PENGWIN, BraTS2024, ATLAS, and Toothfairy2 datasets, respectively, which significantly surpasses existing models.
comment: The corresponding author disagrees with the manuscript submitted to arXiv
♻ ☆ Data Poisoning Attacks to Locally Differentially Private Range Query Protocols
Local Differential Privacy (LDP) has been widely adopted to protect user privacy in decentralized data collection. However, recent studies have revealed that LDP protocols are vulnerable to data poisoning attacks, where malicious users manipulate their reported data to distort aggregated results. In this work, we present the first study on data poisoning attacks targeting LDP range query protocols, focusing on both tree-based and grid-based approaches. We identify three key challenges in executing such attacks, including crafting consistent and effective fake data, maintaining data consistency across levels or grids, and preventing server detection. To address the first two challenges, we propose novel attack methods that are provably optimal, including a tree-based attack and a grid-based attack, designed to manipulate range query results with high effectiveness. \textbf{Our key finding is that the common post-processing procedure, Norm-Sub, in LDP range query protocols can help the attacker massively amplify their attack effectiveness.} In addition, we study a potential countermeasure, but also propose an adaptive attack capable of evading this defense to address the third challenge. We evaluate our methods through theoretical analysis and extensive experiments on synthetic and real-world datasets. Our results show that the proposed attacks can significantly amplify estimations for arbitrary range queries by manipulating a small fraction of users, providing 5-10x more influence than a normal user to the estimation.
♻ ☆ An Efficient Learning Method to Connect Observables
Constructing fast and accurate surrogate models is a key ingredient for making robust predictions in many topics. We introduce a new model, the Multiparameter Eigenvalue Problem (MEP) emulator. The new method connects emulators and can make predictions directly from observables to observables. We present that the MEP emulator can be trained with data from Eigenvector Continuation (EC) and Parametric Matrix Model (PMM) emulators. A simple simulation on a one-dimensional lattice confirms the performance of the MEP emulator. Using $^{28}$O as an example, we also demonstrate that the predictive probability distribution of the target observables can be easily obtained through the new emulator.
comment: 5+2 pages, 4 figures, updated acknowledgment
♻ ☆ Which Frequencies do CNNs Need? Emergent Bottleneck Structure in Feature Learning
We describe the emergence of a Convolution Bottleneck (CBN) structure in CNNs, where the network uses its first few layers to transform the input representation into a representation that is supported only along a few frequencies and channels, before using the last few layers to map back to the outputs. We define the CBN rank, which describes the number and type of frequencies that are kept inside the bottleneck, and partially prove that the parameter norm required to represent a function $f$ scales as depth times the CBN rank $f$. We also show that the parameter norm depends at next order on the regularity of $f$. We show that any network with almost optimal parameter norm will exhibit a CBN structure in both the weights and - under the assumption that the network is stable under large learning rate - the activations, which motivates the common practice of down-sampling; and we verify that the CBN results still hold with down-sampling. Finally we use the CBN structure to interpret the functions learned by CNNs on a number of tasks.
♻ ☆ Hamiltonian Mechanics of Feature Learning: Bottleneck Structure in Leaky ResNets
We study Leaky ResNets, which interpolate between ResNets and Fully-Connected nets depending on an 'effective depth' hyper-parameter $\tilde{L}$. In the infinite depth limit, we study 'representation geodesics' $A_{p}$: continuous paths in representation space (similar to NeuralODEs) from input $p=0$ to output $p=1$ that minimize the parameter norm of the network. We give a Lagrangian and Hamiltonian reformulation, which highlight the importance of two terms: a kinetic energy which favors small layer derivatives $\partial_{p}A_{p}$ and a potential energy that favors low-dimensional representations, as measured by the 'Cost of Identity'. The balance between these two forces offers an intuitive understanding of feature learning in ResNets. We leverage this intuition to explain the emergence of a bottleneck structure, as observed in previous work: for large $\tilde{L}$ the potential energy dominates and leads to a separation of timescales, where the representation jumps rapidly from the high dimensional inputs to a low-dimensional representation, move slowly inside the space of low-dimensional representations, before jumping back to the potentially high-dimensional outputs. Inspired by this phenomenon, we train with an adaptive layer step-size to adapt to the separation of timescales.
♻ ☆ How DNNs break the Curse of Dimensionality: Compositionality and Symmetry Learning
We show that deep neural networks (DNNs) can efficiently learn any composition of functions with bounded $F_{1}$-norm, which allows DNNs to break the curse of dimensionality in ways that shallow networks cannot. More specifically, we derive a generalization bound that combines a covering number argument for compositionality, and the $F_{1}$-norm (or the related Barron norm) for large width adaptivity. We show that the global minimizer of the regularized loss of DNNs can fit for example the composition of two functions $f^{*}=h\circ g$ from a small number of observations, assuming $g$ is smooth/regular and reduces the dimensionality (e.g. $g$ could be the quotient map of the symmetries of $f^{*}$), so that $h$ can be learned in spite of its low regularity. The measures of regularity we consider is the Sobolev norm with different levels of differentiability, which is well adapted to the $F_{1}$ norm. We compute scaling laws empirically and observe phase transitions depending on whether $g$ or $h$ is harder to learn, as predicted by our theory.
♻ ☆ CATCH: Channel-Aware multivariate Time Series Anomaly Detection via Frequency Patching ICLR 2025
Anomaly detection in multivariate time series is challenging as heterogeneous subsequence anomalies may occur. Reconstruction-based methods, which focus on learning normal patterns in the frequency domain to detect diverse abnormal subsequences, achieve promising results, while still falling short on capturing fine-grained frequency characteristics and channel correlations. To contend with the limitations, we introduce CATCH, a framework based on frequency patching. We propose to patchify the frequency domain into frequency bands, which enhances its ability to capture fine-grained frequency characteristics. To perceive appropriate channel correlations, we propose a Channel Fusion Module (CFM), which features a patch-wise mask generator and a masked-attention mechanism. Driven by a bi-level multi-objective optimization algorithm, the CFM is encouraged to iteratively discover appropriate patch-wise channel correlations, and to cluster relevant channels while isolating adverse effects from irrelevant channels. Extensive experiments on 10 real-world datasets and 12 synthetic datasets demonstrate that CATCH achieves state-of-the-art performance. We make our code and datasets available at https://github.com/decisionintelligence/CATCH.
comment: Accepted by ICLR 2025
♻ ☆ AfroBench: How Good are Large Language Models on African Languages?
Large-scale multilingual evaluations, such as MEGA, often include only a handful of African languages due to the scarcity of high-quality evaluation data and the limited discoverability of existing African datasets. This lack of representation hinders comprehensive LLM evaluation across a diverse range of languages and tasks. To address these challenges, we introduce AfroBench -- a multi-task benchmark for evaluating the performance of LLMs across 64 African languages, 15 tasks and 22 datasets. AfroBench consists of nine natural language understanding datasets, six text generation datasets, six knowledge and question answering tasks, and one mathematical reasoning task. We present results comparing the performance of prompting LLMs to fine-tuned baselines based on BERT and T5-style models. Our results suggest large gaps in performance between high-resource languages, such as English, and African languages across most tasks; but performance also varies based on the availability of monolingual data resources. Our findings confirm that performance on African languages continues to remain a hurdle for current LLMs, underscoring the need for additional efforts to close this gap. https://mcgill-nlp.github.io/AfroBench/
comment: Under review
♻ ☆ Robust Amortized Bayesian Inference with Self-Consistency Losses on Unlabeled Data
Neural amortized Bayesian inference (ABI) can solve probabilistic inverse problems orders of magnitude faster than classical methods. However, neural ABI is not yet sufficiently robust for widespread and safe applicability. In particular, when performing inference on observations outside of the scope of the simulated data seen during training, for example, because of model misspecification, the posterior approximations are likely to become highly biased. Due to the bad pre-asymptotic behavior of current neural posterior estimators in the out-of-simulation regime, the resulting estimation biases cannot be fixed in acceptable time by just simulating more training data. In this proof-of-concept paper, we propose a semi-supervised approach that enables training not only on (labeled) simulated data generated from the model, but also on unlabeled data originating from any source, including real-world data. To achieve the latter, we exploit Bayesian self-consistency properties that can be transformed into strictly proper losses without requiring knowledge of true parameter values, that is, without requiring data labels. The results of our initial experiments show remarkable improvements in the robustness of ABI on out-of-simulation data. Even if the observed data is far away from both labeled and unlabeled training data, inference remains highly accurate. If our findings also generalize to other scenarios and model classes, we believe that our new method represents a major breakthrough in neural ABI.
comment: added acknowledgements
♻ ☆ Enhancing Vietnamese VQA through Curriculum Learning on Raw and Augmented Text Representations AAAI-25
Visual Question Answering (VQA) is a multimodal task requiring reasoning across textual and visual inputs, which becomes particularly challenging in low-resource languages like Vietnamese due to linguistic variability and the lack of high-quality datasets. Traditional methods often rely heavily on extensive annotated datasets, computationally expensive pipelines, and large pre-trained models, specifically in the domain of Vietnamese VQA, limiting their applicability in such scenarios. To address these limitations, we propose a training framework that combines a paraphrase-based feature augmentation module with a dynamic curriculum learning strategy. Explicitly, augmented samples are considered "easy" while raw samples are regarded as "hard". The framework then utilizes a mechanism that dynamically adjusts the ratio of easy to hard samples during training, progressively modifying the same dataset to increase its difficulty level. By enabling gradual adaptation to task complexity, this approach helps the Vietnamese VQA model generalize well, thus improving overall performance. Experimental results show consistent improvements on the OpenViVQA dataset and mixed outcomes on the ViVQA dataset, highlighting both the potential and challenges of our approach in advancing VQA for Vietnamese language.
comment: 10 pages, 3 figures, AAAI-25 Workshop on Document Understanding and Intelligence
♻ ☆ Detecting new obfuscated malware variants: A lightweight and interpretable machine learning approach
Machine learning has been successfully applied in developing malware detection systems, with a primary focus on accuracy, and increasing attention to reducing computational overhead and improving model interpretability. However, an important question remains underexplored: How well can machine learning-based models detect entirely new forms of malware not present in the training data? In this study, we present a machine learning-based system for detecting obfuscated malware that is not only highly accurate, lightweight and interpretable, but also capable of successfully adapting to new types of malware attacks. Our system is capable of detecting 15 malware subtypes despite being exclusively trained on one malware subtype, namely the Transponder from the Spyware family. This system was built after training 15 distinct random forest-based models, each on a different malware subtype from the CIC-MalMem-2022 dataset. These models were evaluated against the entire range of malware subtypes, including all unseen malware subtypes. To maintain the system's streamlined nature, training was confined to the top five most important features, which also enhanced interpretability. The Transponder-focused model exhibited high accuracy, exceeding 99.8%, with an average processing speed of 5.7 microseconds per file. We also illustrate how the Shapley additive explanations technique can facilitate the interpretation of the model predictions. Our research contributes to advancing malware detection methodologies, pioneering the feasibility of detecting obfuscated malware by exclusively training a model on a single or a few carefully selected malware subtypes and applying it to detect unseen subtypes.
comment: 30 pages (excluding Appendix), 5 figures and 5 tables. Now published in Intelligent Systems with Applications (https://doi.org/10.1016/j.iswa.2024.200472)
♻ ☆ Nature Language Model: Deciphering the Language of Nature for Scientific Discovery
Foundation models have revolutionized natural language processing and artificial intelligence, significantly enhancing how machines comprehend and generate human languages. Inspired by the success of these foundation models, researchers have developed foundation models for individual scientific domains, including small molecules, materials, proteins, DNA, RNA and even cells. However, these models are typically trained in isolation, lacking the ability to integrate across different scientific domains. Recognizing that entities within these domains can all be represented as sequences, which together form the "language of nature", we introduce Nature Language Model (NatureLM), a sequence-based science foundation model designed for scientific discovery. Pre-trained with data from multiple scientific domains, NatureLM offers a unified, versatile model that enables various applications including: (i) generating and optimizing small molecules, proteins, RNA, and materials using text instructions; (ii) cross-domain generation/design, such as protein-to-molecule and protein-to-RNA generation; and (iii) top performance across different domains, matching or surpassing state-of-the-art specialist models. NatureLM offers a promising generalist approach for various scientific tasks, including drug discovery (hit generation/optimization, ADMET optimization, synthesis), novel material design, and the development of therapeutic proteins or nucleotides. We have developed NatureLM models in different sizes (1 billion, 8 billion, and 46.7 billion parameters) and observed a clear improvement in performance as the model size increases.
comment: 93 pages
♻ ☆ Deep unrolling for learning optimal spatially varying regularisation parameters for Total Generalised Variation
We extend a recently introduced deep unrolling framework for learning spatially varying regularisation parameters in inverse imaging problems to the case of Total Generalised Variation (TGV). The framework combines a deep convolutional neural network (CNN) inferring the two spatially varying TGV parameters with an unrolled algorithmic scheme that solves the corresponding variational problem. The two subnetworks are jointly trained end-to-end in a supervised fashion and as such the CNN learns to compute those parameters that drive the reconstructed images as close to the ground truth as possible. Numerical results in image denoising and MRI reconstruction show a significant qualitative and quantitative improvement compared to the best TGV scalar parameter case as well as to other approaches employing spatially varying parameters computed by unsupervised methods. We also observe that the inferred spatially varying parameter maps have a consistent structure near the image edges, asking for further theoretical investigations. In particular, the parameter that weighs the first-order TGV term has a triple-edge structure with alternating high-low-high values whereas the one that weighs the second-order term attains small values in a large neighbourhood around the edges.
♻ ☆ HelpSteer2-Preference: Complementing Ratings with Preferences ICLR 2025
Reward models are critical for aligning models to follow instructions, and are typically trained following one of two popular paradigms: Bradley-Terry style or Regression style. However, there is a lack of evidence that either approach is better than the other, when adequately matched for data. This is primarily because these approaches require data collected in different (but incompatible) formats, meaning that adequately matched data is not available in existing public datasets. To tackle this problem, we release preference annotations (designed for Bradley-Terry training) to complement existing ratings (designed for Regression style training) in the HelpSteer2 dataset. To improve data interpretability, preference annotations are accompanied with human-written justifications. Using this data, we conduct the first head-to-head comparison of Bradley-Terry and Regression models when adequately matched for data. Based on insights derived from such a comparison, we propose a novel approach to combine Bradley-Terry and Regression reward modeling. A Llama-3.1-70B-Instruct model tuned with this approach scores 94.1 on RewardBench, emerging top of more than 140 reward models as of 1 Oct 2024. This reward model can then be used with REINFORCE algorithm (RLHF) to align an Instruct model to reach 85.0 on Arena Hard, which is No. 1 as of 1 Oct 2024. We open-source this dataset (CC-BY-4.0 license) at https://huggingface.co/datasets/nvidia/HelpSteer2#preferences-new -- 1-oct-2024 and openly release the trained Reward and Instruct models at https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Reward and https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct
comment: Accepted to ICLR 2025; 28 pages, 3 figures
♻ ☆ Cross-Modal Prototype based Multimodal Federated Learning under Severely Missing Modality
Multimodal federated learning (MFL) has emerged as a decentralized machine learning paradigm, allowing multiple clients with different modalities to collaborate on training a global model across diverse data sources without sharing their private data. However, challenges, such as data heterogeneity and severely missing modalities, pose crucial hindrances to the robustness of MFL, significantly impacting the performance of global model. The occurrence of missing modalities in real-world applications, such as autonomous driving, often arises from factors like sensor failures, leading knowledge gaps during the training process. Specifically, the absence of a modality introduces misalignment during the local training phase, stemming from zero-filling in the case of clients with missing modalities. Consequently, achieving robust generalization in global model becomes imperative, especially when dealing with clients that have incomplete data. In this paper, we propose $\textbf{Multimodal Federated Cross Prototype Learning (MFCPL)}$, a novel approach for MFL under severely missing modalities. Our MFCPL leverages the complete prototypes to provide diverse modality knowledge in modality-shared level with the cross-modal regularization and modality-specific level with cross-modal contrastive mechanism. Additionally, our approach introduces the cross-modal alignment to provide regularization for modality-specific features, thereby enhancing the overall performance, particularly in scenarios involving severely missing modalities. Through extensive experiments on three multimodal datasets, we demonstrate the effectiveness of MFCPL in mitigating the challenges of data heterogeneity and severely missing modalities while improving the overall performance and robustness of MFL.
comment: 14 pages, 8 figures, 11 tables
♻ ☆ Extracting Formulae in Many-Valued Logic from Deep Neural Networks
We propose a new perspective on deep ReLU networks, namely as circuit counterparts of Lukasiewicz infinite-valued logic -- a many-valued (MV) generalization of Boolean logic. An algorithm for extracting formulae in MV logic from deep ReLU networks is presented. As the algorithm applies to networks with general, in particular also real-valued, weights, it can be used to extract logical formulae from deep ReLU networks trained on data.
comment: Signicant extension of the previous version
♻ ☆ Learning truly monotone operators with applications to nonlinear inverse problems
This article introduces a novel approach to learning monotone neural networks through a newly defined penalization loss. The proposed method is particularly effective in solving classes of variational problems, specifically monotone inclusion problems, commonly encountered in image processing tasks. The Forward-Backward-Forward (FBF) algorithm is employed to address these problems, offering a solution even when the Lipschitz constant of the neural network is unknown. Notably, the FBF algorithm provides convergence guarantees under the condition that the learned operator is monotone. Building on plug-and-play methodologies, our objective is to apply these newly learned operators to solving non-linear inverse problems. To achieve this, we initially formulate the problem as a variational inclusion problem. Subsequently, we train a monotone neural network to approximate an operator that may not inherently be monotone. Leveraging the FBF algorithm, we then show simulation examples where the non-linear inverse problem is successfully solved.
♻ ☆ A conversion theorem and minimax optimality for continuum contextual bandits
We study the contextual continuum bandits problem, where the learner sequentially receives a side information vector and has to choose an action in a convex set, minimizing a function associated with the context. The goal is to minimize all the underlying functions for the received contexts, leading to the contextual notion of regret, which is stronger than the standard static regret. Assuming that the objective functions are $\gamma$-H\"older with respect to the contexts, $0<\gamma\le 1,$ we demonstrate that any algorithm achieving a sub-linear static regret can be extended to achieve a sub-linear contextual regret. We prove a static-to-contextual regret conversion theorem that provides an upper bound for the contextual regret of the output algorithm as a function of the static regret of the input algorithm. We further study the implications of this general result for three fundamental cases of dependency of the objective function on the action variable: (a) Lipschitz bandits, (b) convex bandits, (c) strongly convex and smooth bandits. For Lipschitz bandits and $\gamma=1,$ combining our results with the lower bound of Slivkins (2014), we prove that the minimax optimal contextual regret for the noise-free adversarial setting is achieved. Then, we prove that in the presence of noise, the contextual regret rate as a function of the number of queries is the same for convex bandits as it is for strongly convex and smooth bandits. Lastly, we present a minimax lower bound, implying two key facts. First, obtaining a sub-linear contextual regret may be impossible over functions that are not continuous with respect to the context. Second, for convex bandits and strongly convex and smooth bandits, the algorithms that we propose achieve, up to a logarithmic factor, the minimax optimal rate of contextual regret as a function of the number of queries.
♻ ☆ Federated Learning With Individualized Privacy Through Client Sampling ICML
With growing concerns about user data collection, individualized privacy has emerged as a promising solution to balance protection and utility by accounting for diverse user privacy preferences. Instead of enforcing a uniform level of anonymization for all users, this approach allows individuals to choose privacy settings that align with their comfort levels. Building on this idea, we propose an adapted method for enabling Individualized Differential Privacy (IDP) in Federated Learning (FL) by handling clients according to their personal privacy preferences. By extending the SAMPLE algorithm from centralized settings to FL, we calculate client-specific sampling rates based on their heterogeneous privacy budgets and integrate them into a modified IDP-FedAvg algorithm. We test this method under realistic privacy distributions and multiple datasets. The experimental results demonstrate that our approach achieves clear improvements over uniform DP baselines, reducing the trade-off between privacy and utility. Compared to the alternative SCALE method in related work, which assigns differing noise scales to clients, our method performs notably better. However, challenges remain for complex tasks with non-i.i.d. data, primarily stemming from the constraints of the decentralized setting.
comment: Accepted at 10th International Conference on Machine Learning Technologies (ICMLT 2025)
♻ ☆ Decentralized Sporadic Federated Learning: A Unified Algorithmic Framework with Convergence Guarantees
Decentralized federated learning (DFL) captures FL settings where both (i) model updates and (ii) model aggregations are exclusively carried out by the clients without a central server. Existing DFL works have mostly focused on settings where clients conduct a fixed number of local updates between local model exchanges, overlooking heterogeneity and dynamics in communication and computation capabilities. In this work, we propose Decentralized Sporadic Federated Learning ($\texttt{DSpodFL}$), a DFL methodology built on a generalized notion of $\textit{sporadicity}$ in both local gradient and aggregation processes. $\texttt{DSpodFL}$ subsumes many existing decentralized optimization methods under a unified algorithmic framework by modeling the per-iteration (i) occurrence of gradient descent at each client and (ii) exchange of models between client pairs as arbitrary indicator random variables, thus capturing $\textit{heterogeneous and time-varying}$ computation/communication scenarios. We analytically characterize the convergence behavior of $\texttt{DSpodFL}$ for both convex and non-convex models and for both constant and diminishing learning rates, under mild assumptions on the communication graph connectivity, data heterogeneity across clients, and gradient noises. We show how our bounds recover existing results from decentralized gradient descent as special cases. Experiments demonstrate that $\texttt{DSpodFL}$ consistently achieves improved training speeds compared with baselines under various system settings.
♻ ☆ Social Genome: Grounded Social Reasoning Abilities of Multimodal Models
Social reasoning abilities are crucial for AI systems to effectively interpret and respond to multimodal human communication and interaction within social contexts. We introduce Social Genome, the first benchmark for fine-grained, grounded social reasoning abilities of multimodal models. Social Genome contains 272 videos of interactions and 1,486 human-annotated reasoning traces related to inferences about these interactions. These traces contain 5,777 reasoning steps that reference evidence from visual cues, verbal cues, vocal cues, and external knowledge (contextual knowledge external to videos). Social Genome is also the first modeling challenge to study external knowledge in social reasoning. Social Genome computes metrics to holistically evaluate semantic and structural qualities of model-generated social reasoning traces. We demonstrate the utility of Social Genome through experiments with state-of-the-art models, identifying performance gaps and opportunities for future research to improve the grounded social reasoning abilities of multimodal models.
comment: Under Review, 22 pages
♻ ☆ $σ$-zero: Gradient-based Optimization of $\ell_0$-norm Adversarial Examples ICLR 2025
Evaluating the adversarial robustness of deep networks to gradient-based attacks is challenging. While most attacks consider $\ell_2$- and $\ell_\infty$-norm constraints to craft input perturbations, only a few investigate sparse $\ell_1$- and $\ell_0$-norm attacks. In particular, $\ell_0$-norm attacks remain the least studied due to the inherent complexity of optimizing over a non-convex and non-differentiable constraint. However, evaluating adversarial robustness under these attacks could reveal weaknesses otherwise left untested with more conventional $\ell_2$- and $\ell_\infty$-norm attacks. In this work, we propose a novel $\ell_0$-norm attack, called $\sigma$-zero, which leverages a differentiable approximation of the $\ell_0$ norm to facilitate gradient-based optimization, and an adaptive projection operator to dynamically adjust the trade-off between loss minimization and perturbation sparsity. Extensive evaluations using MNIST, CIFAR10, and ImageNet datasets, involving robust and non-robust models, show that $\sigma$\texttt{-zero} finds minimum $\ell_0$-norm adversarial examples without requiring any time-consuming hyperparameter tuning, and that it outperforms all competing sparse attacks in terms of success rate, perturbation size, and efficiency.
comment: Paper accepted at International Conference on Learning Representations (ICLR 2025). Code available at https://github.com/sigma0-advx/sigma-zero
♻ ☆ VISION-XL: High Definition Video Inverse Problem Solver using Latent Image Diffusion Models
In this paper, we propose a novel framework for solving high-definition video inverse problems using latent image diffusion models. Building on recent advancements in spatio-temporal optimization for video inverse problems using image diffusion models, our approach leverages latent-space diffusion models to achieve enhanced video quality and resolution. To address the high computational demands of processing high-resolution frames, we introduce a pseudo-batch consistent sampling strategy, allowing efficient operation on a single GPU. Additionally, to improve temporal consistency, we present pseudo-batch inversion, an initialization technique that incorporates informative latents from the measurement. By integrating with SDXL, our framework achieves state-of-the-art video reconstruction across a wide range of spatio-temporal inverse problems, including complex combinations of frame averaging and various spatial degradations, such as deblurring, super-resolution, and inpainting. Unlike previous methods, our approach supports multiple aspect ratios (landscape, vertical, and square) and delivers HD-resolution reconstructions (exceeding 1280x720) in under 6 seconds per frame on a single NVIDIA 4090 GPU.
comment: Project page: https://vision-xl.github.io/
♻ ☆ No More Sliding Window: Efficient 3D Medical Image Segmentation with Differentiable Top-k Patch Sampling
3D models surpass 2D models in CT/MRI segmentation by effectively capturing inter-slice relationships. However, the added depth dimension substantially increases memory consumption. While patch-based training alleviates memory constraints, it significantly slows down the inference speed due to the sliding window (SW) approach. We propose No-More-Sliding-Window (NMSW), a novel end-to-end trainable framework that enhances the efficiency of generic 3D segmentation backbone during an inference step by eliminating the need for SW. NMSW employs a differentiable Top-k module to selectively sample only the most relevant patches, thereby minimizing redundant computations. When patch-level predictions are insufficient, the framework intelligently leverages coarse global predictions to refine results. Evaluated across 3 tasks using 3 segmentation backbones, NMSW achieves competitive accuracy compared to SW inference while significantly reducing computational complexity by 91% (88.0 to 8.00 TMACs). Moreover, it delivers a 9.1x faster inference on the H100 GPU (99.0 to 8.3 sec) and a 11.1x faster inference on the Xeon Gold CPU (2110 to 189 sec). NMSW is model-agnostic, further boosting efficiency when integrated with any existing efficient segmentation backbones.
♻ ☆ Unveiling the Power of Noise Priors: Enhancing Diffusion Models for Mobile Traffic Prediction
Accurate prediction of mobile traffic, \textit{i.e.,} network traffic from cellular base stations, is crucial for optimizing network performance and supporting urban development. However, the non-stationary nature of mobile traffic, driven by human activity and environmental changes, leads to both regular patterns and abrupt variations. Diffusion models excel in capturing such complex temporal dynamics due to their ability to capture the inherent uncertainties. Most existing approaches prioritize designing novel denoising networks but often neglect the critical role of noise itself, potentially leading to sub-optimal performance. In this paper, we introduce a novel perspective by emphasizing the role of noise in the denoising process. Our analysis reveals that noise fundamentally shapes mobile traffic predictions, exhibiting distinct and consistent patterns. We propose NPDiff, a framework that decomposes noise into \textit{prior} and \textit{residual} components, with the \textit{prior} derived from data dynamics, enhancing the model's ability to capture both regular and abrupt variations. NPDiff can seamlessly integrate with various diffusion-based prediction models, delivering predictions that are effective, efficient, and robust. Extensive experiments demonstrate that it achieves superior performance with an improvement over 30\%, offering a new perspective on leveraging diffusion models in this domain.
♻ ☆ KAGNNs: Kolmogorov-Arnold Networks meet Graph Learning
In recent years, Graph Neural Networks (GNNs) have become the de facto tool for learning node and graph representations. Most GNNs typically consist of a sequence of neighborhood aggregation (a.k.a., message-passing) layers, within which the representation of each node is updated based on those of its neighbors. The most expressive message-passing GNNs can be obtained through the use of the sum aggregator and of MLPs for feature transformation, thanks to their universal approximation capabilities. However, the limitations of MLPs recently motivated the introduction of another family of universal approximators, called Kolmogorov-Arnold Networks (KANs) which rely on a different representation theorem. In this work, we compare the performance of KANs against that of MLPs on graph learning tasks. We implement three new KAN-based GNN layers, inspired respectively by the GCN, GAT and GIN layers. We evaluate two different implementations of KANs using two distinct base families of functions, namely B-splines and radial basis functions. We perform extensive experiments on node classification, link prediction, graph classification and graph regression datasets. Our results indicate that KANs are on-par with or better than MLPs on all tasks studied in this paper. We also show that the size and training speed of RBF-based KANs is only marginally higher than for MLPs, making them viable alternatives. Code available at https://github.com/RomanBresson/KAGNN.
♻ ☆ Graph Neural Networks for Edge Signals: Orientation Equivariance and Invariance
Many applications in traffic, civil engineering, or electrical engineering revolve around edge-level signals. Such signals can be categorized as inherently directed, for example, the water flow in a pipe network, and undirected, like the diameter of a pipe. Topological methods model edge signals with inherent direction by representing them relative to a so-called orientation assigned to each edge. These approaches can neither model undirected edge signals nor distinguish if an edge itself is directed or undirected. We address these shortcomings by (i) revising the notion of orientation equivariance to enable edge direction-aware topological models, (ii) proposing orientation invariance as an additional requirement to describe signals without inherent direction, and (iii) developing EIGN, an architecture composed of novel direction-aware edge-level graph shift operators, that provably fulfills the aforementioned desiderata. It is the first general-purpose topological GNN for edge-level signals that can model directed and undirected signals while distinguishing between directed and undirected edges. A comprehensive evaluation shows that EIGN outperforms prior work in edge-level tasks, for example, improving in RMSE on flow simulation tasks by up to 23.5%.
♻ ☆ GIFT: Unlocking Full Potential of Labels in Distilled Dataset at Near-zero Cost
Recent advancements in dataset distillation have demonstrated the significant benefits of employing soft labels generated by pre-trained teacher models. In this paper, we introduce a novel perspective by emphasizing the full utilization of labels. We first conduct a comprehensive comparison of various loss functions for soft label utilization in dataset distillation, revealing that the model trained on the synthetic dataset exhibits high sensitivity to the choice of loss function for soft label utilization. This finding highlights the necessity of a universal loss function for training models on synthetic datasets. Building on these insights, we introduce an extremely simple yet surprisingly effective plug-and-play approach, GIFT, which encompasses soft label refinement and a cosine similarity-based loss function to efficiently leverage full label information. Extensive experiments indicate that GIFT consistently enhances state-of-the-art dataset distillation methods across various dataset scales, without incurring additional computational costs. Importantly, GIFT significantly enhances cross-optimizer generalization, an area previously overlooked. For instance, on ImageNet-1K with IPC = 10, GIFT enhances the state-of-the-art method RDED by 30.8% in cross-optimizer generalization. Our code is available at https://github.com/LINs-lab/GIFT.
comment: https://github.com/LINs-lab/GIFT
♻ ☆ Estimation of multiple mean vectors in high dimension
We endeavour to estimate numerous multi-dimensional means of various probability distributions on a common space based on independent samples. Our approach involves forming estimators through convex combinations of empirical means derived from these samples. We introduce two strategies to find appropriate data-dependent convex combination weights: a first one employing a testing procedure to identify neighbouring means with low variance, which results in a closed-form plug-in formula for the weights, and a second one determining weights via minimization of an upper confidence bound on the quadratic risk.Through theoretical analysis, we evaluate the improvement in quadratic risk offered by our methods compared to the empirical means. Our analysis focuses on a dimensional asymptotics perspective, showing that our methods asymptotically approach an oracle (minimax) improvement as the effective dimension of the data increases.We demonstrate the efficacy of our methods in estimating multiple kernel mean embeddings through experiments on both simulated and real-world datasets.
♻ ☆ Assessing Pre-Trained Models for Transfer Learning Through Distribution of Spectral Components AAAI 2025
Pre-trained model assessment for transfer learning aims to identify the optimal candidate for the downstream tasks from a model hub, without the need of time-consuming fine-tuning. Existing advanced works mainly focus on analyzing the intrinsic characteristics of the entire features extracted by each pre-trained model or how well such features fit the target labels. This paper proposes a novel perspective for pre-trained model assessment through the Distribution of Spectral Components (DISCO). Through singular value decomposition of features extracted from pre-trained models, we investigate different spectral components and observe that they possess distinct transferability, contributing diversely to the fine-tuning performance. Inspired by this, we propose an assessment method based on the distribution of spectral components which measures the proportions of their corresponding singular values. Pre-trained models with features concentrating on more transferable components are regarded as better choices for transfer learning. We further leverage the labels of downstream data to better estimate the transferability of each spectral component and derive the final assessment criterion. Our proposed method is flexible and can be applied to both classification and regression tasks. We conducted comprehensive experiments across three benchmarks and two tasks including image classification and object detection, demonstrating that our method achieves state-of-the-art performance in choosing proper pre-trained models from the model hub for transfer learning.
comment: Accepted by AAAI 2025
♻ ☆ Reinforcement Learning on Reconfigurable Hardware: Overcoming Material Variability in Laser Material Processing ICRA
Ensuring consistent processing quality is challenging in laser processes due to varying material properties and surface conditions. Although some approaches have shown promise in solving this problem via automation, they often rely on predetermined targets or are limited to simulated environments. To address these shortcomings, we propose a novel real-time reinforcement learning approach for laser process control, implemented on a Field Programmable Gate Array to achieve real-time execution. Our experimental results from laser welding tests on stainless steel samples with a range of surface roughnesses validated the method's ability to adapt autonomously, without relying on reward engineering or prior setup information. Specifically, the algorithm learned the correct power profile for each unique surface characteristic, demonstrating significant improvements over hand-engineered optimal constant power strategies -- up to 23% better performance on rougher surfaces and 7% on mixed surfaces. This approach represents a significant advancement in automating and optimizing laser processes, with potential applications across multiple industries.
comment: Accepted for the 2025 IEEE International Conference on Robotics and Automation (ICRA), May 19-23, 2025, Atlanta, USA; Camera ready version -- addressed reviewer comments in text, improved plot clarity
♻ ☆ Wasserstein-regularized Conformal Prediction under General Distribution Shift
Conformal prediction yields a prediction set with guaranteed $1-\alpha$ coverage of the true target under the i.i.d. assumption, which may not hold and lead to a gap between $1-\alpha$ and the actual coverage. Prior studies bound the gap using total variation distance, which cannot identify the gap changes under distribution shift at a given $\alpha$. Besides, existing methods are mostly limited to covariate shift,while general joint distribution shifts are more common in practice but less researched.In response, we first propose a Wasserstein distance-based upper bound of the coverage gap and analyze the bound using probability measure pushforwards between the shifted joint data and conformal score distributions, enabling a separation of the effect of covariate and concept shifts over the coverage gap. We exploit the separation to design an algorithm based on importance weighting and regularized representation learning (WR-CP) to reduce the Wasserstein bound with a finite-sample error bound.WR-CP achieves a controllable balance between conformal prediction accuracy and efficiency. Experiments on six datasets prove that WR-CP can reduce coverage gaps to $3.2\%$ across different confidence levels and outputs prediction sets 37$\%$ smaller than the worst-case approach on average.
♻ ☆ Mixed Graph Contrastive Network for Semi-Supervised Node Classification
Graph Neural Networks (GNNs) have achieved promising performance in semi-supervised node classification in recent years. However, the problem of insufficient supervision, together with representation collapse, largely limits the performance of the GNNs in this field. To alleviate the collapse of node representations in semi-supervised scenario, we propose a novel graph contrastive learning method, termed Mixed Graph Contrastive Network (MGCN). In our method, we improve the discriminative capability of the latent embeddings by an interpolation-based augmentation strategy and a correlation reduction mechanism. Specifically, we first conduct the interpolation-based augmentation in the latent space and then force the prediction model to change linearly between samples. Second, we enable the learned network to tell apart samples across two interpolation-perturbed views through forcing the correlation matrix across views to approximate an identity matrix. By combining the two settings, we extract rich supervision information from both the abundant unlabeled nodes and the rare yet valuable labeled nodes for discriminative representation learning. Extensive experimental results on six datasets demonstrate the effectiveness and the generality of MGCN compared to the existing state-of-the-art methods. The code of MGCN is available at https://github.com/xihongyang1999/MGCN on Github.
♻ ☆ Rethinking Weight-Averaged Model-merging
Model-merging has emerged as a powerful approach in deep learning, capable of enhancing model performance without any training. However, the underlying mechanisms that explain its effectiveness remain largely unexplored. In this paper, we investigate this technique from three novel perspectives to empirically provide deeper insights into why and how weight-averaged model-merging works: (1) we examine the intrinsic patterns captured by the learning of the model weights, through the visualizations of their patterns on several datasets, showing that these weights often encode structured and interpretable patterns and that is the essential why model-merging can work; (2) we mathematically and empirically investigate model ensemble merging strategies based on averaging on weights versus averaging on features, providing detailed analyses across diverse architectures and datasets; and (3) we explore the impact on model-merging prediction stability in terms of changing the parameter magnitude, revealing insights into the way of weight averaging works as regularization by showing the robustness across different parameter scales. Our findings shed light on the "black box" of weight-averaged model-merging, offering valuable insights and practical recommendations that advance the model-merging process. The code is available at https://github.com/billhhh/Rethink-Merge.
♻ ☆ Careful with that Scalpel: Improving Gradient Surgery with an EMA
Beyond minimizing a single training loss, many deep learning estimation pipelines rely on an auxiliary objective to quantify and encourage desirable properties of the model (e.g. performance on another dataset, robustness, agreement with a prior). Although the simplest approach to incorporating an auxiliary loss is to sum it with the training loss as a regularizer, recent works have shown that one can improve performance by blending the gradients beyond a simple sum; this is known as gradient surgery. We cast the problem as a constrained minimization problem where the auxiliary objective is minimized among the set of minimizers of the training loss. To solve this bilevel problem, we follow a parameter update direction that combines the training loss gradient and the orthogonal projection of the auxiliary gradient to the training gradient. In a setting where gradients come from mini-batches, we explain how, using a moving average of the training loss gradients, we can carefully maintain this critical orthogonality property. We demonstrate that our method, Bloop, can lead to much better performances on NLP and vision experiments than other gradient surgery methods without EMA.
♻ ☆ Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed Transformer
We propose Union-of-Experts (UoE), which decomposes transformer into an equitant group of experts, and then implement selective routing on input data and experts. Our approach advances MoE design with four key innovations: (1) We conducted equitant expert decomposition on both MLP blocks and attention blocks based on matrix partition in tensor parallelism. (2) We developed two routing paradigms: patch-wise data selection and expert selection, to apply routing across different levels. (3) We design the architecture of UoE model, including Selective Multi-Head Attention (SMHA) and Union-of-MLP-Experts (UoME). (4) We develop parallel implementation of UoE's routing and computation operation, and optimize efficiency based on the hardware processing analysis. The experiments demonstrate that the UoE model surpass Full Attention, state-of-art MoEs and efficient transformers (including the model architecture of recently proposed DeepSeek-V3) in several tasks across image and natural language domains. In language modeling tasks, we achieve an average reduction of 2.38 in perplexity compared to the best-performed MoE method with an average of 76% FLOPs. In Long Range Arena benchmark, we recorded an average score that is at least 0.68% higher than all comparison models including Full Attention, MoEs, and transformer variants, with only 50% FLOPs of the best MoE method. In image classification, our model yielded an average accuracy improvement of 1.75% than the best model while maintaining comparable FLOPs. The source codes are available at https://github.com/YujiaoYang-work/UoE.
comment: 17 pages
♻ ☆ CE-U: Cross Entropy Unlearning
Large language models (LLMs) inadvertently memorize sensitive data from their massive pretraining corpora \cite{jang2022knowledge}. In this work, we propose CE-U (Cross Entropy Unlearning), a novel loss function designed specifically for unlearning tasks. CE-U addresses fundamental limitations of gradient ascent approaches which suffer from instability due to vanishing gradients when model confidence is high and gradient exploding when confidence is low. We also unify standard cross entropy supervision and cross entropy unlearning into a single framework. Notably, on the TOFU benchmark for unlearning \cite{maini2024tofu}, CE-U achieves state-of-the-art results on LLaMA2-7B with 1\% and 5\% forgetting, even without the use of any extra reference model or additional positive samples. Our theoretical analysis further reveals that the gradient instability issues also exist in popular reinforcement learning algorithms like DPO \cite{rafailov2023direct} and GRPO\cite{Shao2024DeepSeekMath}, as they include a gradient ascent component. This suggests that applying CE-U principles to reinforcement learning could be a promising direction for improving stability and convergence.
♻ ☆ Comparing hundreds of machine learning classifiers and discrete choice models in predicting travel behavior: an empirical benchmark
Numerous studies have compared machine learning (ML) and discrete choice models (DCMs) in predicting travel demand. However, these studies often lack generalizability as they compare models deterministically without considering contextual variations. To address this limitation, our study develops an empirical benchmark by designing a tournament model, thus efficiently summarizing a large number of experiments, quantifying the randomness in model comparisons, and using formal statistical tests to differentiate between the model and contextual effects. This benchmark study compares two large-scale data sources: a database compiled from literature review summarizing 136 experiments from 35 studies, and our own experiment data, encompassing a total of 6,970 experiments from 105 models and 12 model families. This benchmark study yields two key findings. Firstly, many ML models, particularly the ensemble methods and deep learning, statistically outperform the DCM family (i.e., multinomial, nested, and mixed logit models). However, this study also highlights the crucial role of the contextual factors (i.e., data sources, inputs and choice categories), which can explain models' predictive performance more effectively than the differences in model types alone. Model performance varies significantly with data sources, improving with larger sample sizes and lower dimensional alternative sets. After controlling all the model and contextual factors, significant randomness still remains, implying inherent uncertainty in such model comparisons. Overall, we suggest that future researchers shift more focus from context-specific model comparisons towards examining model transferability across contexts and characterizing the inherent uncertainty in ML, thus creating more robust and generalizable next-generation travel demand models.
Information Retrieval 17
☆ RadIR: A Scalable Framework for Multi-Grained Medical Image Retrieval via Radiology Report Mining
Developing advanced medical imaging retrieval systems is challenging due to the varying definitions of `similar images' across different medical contexts. This challenge is compounded by the lack of large-scale, high-quality medical imaging retrieval datasets and benchmarks. In this paper, we propose a novel methodology that leverages dense radiology reports to define image-wise similarity ordering at multiple granularities in a scalable and fully automatic manner. Using this approach, we construct two comprehensive medical imaging retrieval datasets: MIMIC-IR for Chest X-rays and CTRATE-IR for CT scans, providing detailed image-image ranking annotations conditioned on diverse anatomical structures. Furthermore, we develop two retrieval systems, RadIR-CXR and model-ChestCT, which demonstrate superior performance in traditional image-image and image-report retrieval tasks. These systems also enable flexible, effective image retrieval conditioned on specific anatomical structures described in text, achieving state-of-the-art results on 77 out of 78 metrics.
☆ IFIR: A Comprehensive Benchmark for Evaluating Instruction-Following in Expert-Domain Information Retrieval NAACL 2025
We introduce IFIR, the first comprehensive benchmark designed to evaluate instruction-following information retrieval (IR) in expert domains. IFIR includes 2,426 high-quality examples and covers eight subsets across four specialized domains: finance, law, healthcare, and science literature. Each subset addresses one or more domain-specific retrieval tasks, replicating real-world scenarios where customized instructions are critical. IFIR enables a detailed analysis of instruction-following retrieval capabilities by incorporating instructions at different levels of complexity. We also propose a novel LLM-based evaluation method to provide a more precise and reliable assessment of model performance in following instructions. Through extensive experiments on 15 frontier retrieval models, including those based on LLMs, our results reveal that current models face significant challenges in effectively following complex, domain-specific instructions. We further provide in-depth analyses to highlight these limitations, offering valuable insights to guide future advancements in retriever development.
comment: NAACL 2025 Main
☆ Training-Free Graph Filtering via Multimodal Feature Refinement for Extremely Fast Multimodal Recommendation
Multimodal recommender systems improve the performance of canonical recommender systems with no item features by utilizing diverse content types such as text, images, and videos, while alleviating inherent sparsity of user-item interactions and accelerating user engagement. However, current neural network-based models often incur significant computational overhead due to the complex training process required to learn and integrate information from multiple modalities. To overcome this limitation, we propose MultiModal-Graph Filtering (MM-GF), a training-free method based on the notion of graph filtering (GF) for efficient and accurate multimodal recommendations. Specifically, MM-GF first constructs multiple similarity graphs through nontrivial multimodal feature refinement such as robust scaling and vector shifting by addressing the heterogeneous characteristics across modalities. Then, MM-GF optimally fuses multimodal information using linear low-pass filters across different modalities. Extensive experiments on real-world benchmark datasets demonstrate that MM-GF not only improves recommendation accuracy by up to 13.35% compared to the best competitor but also dramatically reduces computational costs by achieving the runtime of less than 10 seconds.
comment: 10 pages, 6 figures, 6 tables
☆ In-depth Analysis of Graph-based RAG in a Unified Framework
Graph-based Retrieval-Augmented Generation (RAG) has proven effective in integrating external knowledge into large language models (LLMs), improving their factual accuracy, adaptability, interpretability, and trustworthiness. A number of graph-based RAG methods have been proposed in the literature. However, these methods have not been systematically and comprehensively compared under the same experimental settings. In this paper, we first summarize a unified framework to incorporate all graph-based RAG methods from a high-level perspective. We then extensively compare representative graph-based RAG methods over a range of questing-answering (QA) datasets -- from specific questions to abstract questions -- and examine the effectiveness of all methods, providing a thorough analysis of graph-based RAG approaches. As a byproduct of our experimental analysis, we are also able to identify new variants of the graph-based RAG methods over specific QA and abstract QA tasks respectively, by combining existing techniques, which outperform the state-of-the-art methods. Finally, based on these findings, we offer promising research opportunities. We believe that a deeper understanding of the behavior of existing methods can provide new valuable insights for future research.
☆ Measuring temporal effects of agent knowledge by date-controlled tool use
Temporal progression is an integral part of knowledge accumulation and update. Web search is frequently adopted as grounding for agent knowledge, yet its inappropriate configuration affects the quality of agent responses. Here, we construct a tool-based out-of-sample testing framework to measure the knowledge variability of large language model (LLM) agents from distinct date-controlled tools (DCTs). We demonstrate the temporal effects of an LLM agent as a writing assistant, which can use web search to help complete scientific publication abstracts. We show that temporal effects of the search engine translates into tool-dependent agent performance but can be alleviated with base model choice and explicit reasoning instructions such as chain-of-thought prompting. Our results indicate that agent evaluation should take a dynamical view and account for the temporal influence of tools and the updates of external resources.
comment: comments welcome
☆ Semantic Retrieval Augmented Contrastive Learning for Sequential Recommendation
Sequential recommendation aims to model user preferences based on historical behavior sequences, which is crucial for various online platforms. Data sparsity remains a significant challenge in this area as most users have limited interactions and many items receive little attention. To mitigate this issue, contrastive learning has been widely adopted. By constructing positive sample pairs from the data itself and maximizing their agreement in the embedding space,it can leverage available data more effectively. Constructing reasonable positive sample pairs is crucial for the success of contrastive learning. However, current approaches struggle to generate reliable positive pairs as they either rely on representations learned from inherently sparse collaborative signals or use random perturbations which introduce significant uncertainty. To address these limitations, we propose a novel approach named Semantic Retrieval Augmented Contrastive Learning (SRA-CL), which leverages semantic information to improve the reliability of contrastive samples. SRA-CL comprises two main components: (1) Cross-Sequence Contrastive Learning via User Semantic Retrieval, which utilizes large language models (LLMs) to understand diverse user preferences and retrieve semantically similar users to form reliable positive samples through a learnable sample synthesis method; and (2) Intra-Sequence Contrastive Learning via Item Semantic Retrieval, which employs LLMs to comprehend items and retrieve similar items to perform semantic-based item substitution, thereby creating semantically consistent augmented views for contrastive learning. SRA-CL is plug-and-play and can be integrated into standard sequential recommendation models. Extensive experiments on four public datasets demonstrate the effectiveness and generalizability of the proposed approach.
☆ HEISIR: Hierarchical Expansion of Inverted Semantic Indexing for Training-free Retrieval of Conversational Data using LLMs NAACL 2025
The growth of conversational AI services has increased demand for effective information retrieval from dialogue data. However, existing methods often face challenges in capturing semantic intent or require extensive labeling and fine-tuning. This paper introduces HEISIR (Hierarchical Expansion of Inverted Semantic Indexing for Retrieval), a novel framework that enhances semantic understanding in conversational data retrieval through optimized data ingestion, eliminating the need for resource-intensive labeling or model adaptation. HEISIR implements a two-step process: (1) Hierarchical Triplets Formulation and (2) Adjunct Augmentation, creating semantic indices consisting of Subject-Verb-Object-Adjunct (SVOA) quadruplets. This structured representation effectively captures the underlying semantic information from dialogue content. HEISIR achieves high retrieval performance while maintaining low latency during the actual retrieval process. Our experimental results demonstrate that HEISIR outperforms fine-tuned models across various embedding types and language models. Beyond improving retrieval capabilities, HEISIR also offers opportunities for intent and topic analysis in conversational data, providing a versatile solution for dialogue systems.
comment: Accepted by NAACL 2025 (Findings)
☆ Dynamic-KGQA: A Scalable Framework for Generating Adaptive Question Answering Datasets
As question answering (QA) systems advance alongside the rapid evolution of foundation models, the need for robust, adaptable, and large-scale evaluation benchmarks becomes increasingly critical. Traditional QA benchmarks are often static and publicly available, making them susceptible to data contamination and memorization by large language models (LLMs). Consequently, static benchmarks may overestimate model generalization and hinder a reliable assessment of real-world performance. In this work, we introduce Dynamic-KGQA, a scalable framework for generating adaptive QA datasets from knowledge graphs (KGs), designed to mitigate memorization risks while maintaining statistical consistency across iterations. Unlike fixed benchmarks, Dynamic-KGQA generates a new dataset variant on every run while preserving the underlying distribution, enabling fair and reproducible evaluations. Furthermore, our framework provides fine-grained control over dataset characteristics, supporting domain-specific and topic-focused QA dataset generation. Additionally, Dynamic-KGQA produces compact, semantically coherent subgraphs that facilitate both training and evaluation of KGQA models, enhancing their ability to leverage structured knowledge effectively. To align with existing evaluation protocols, we also provide static large-scale train/test/validation splits, ensuring comparability with prior methods. By introducing a dynamic, customizable benchmarking paradigm, Dynamic-KGQA enables a more rigorous and adaptable evaluation of QA systems.
☆ Collapse of Dense Retrievers: Short, Early, and Literal Biases Outranking Factual Evidence
Dense retrieval models are commonly used in Information Retrieval (IR) applications, such as Retrieval-Augmented Generation (RAG). Since they often serve as the first step in these systems, their robustness is critical to avoid failures. In this work, by repurposing a relation extraction dataset (e.g. Re-DocRED), we design controlled experiments to quantify the impact of heuristic biases, such as favoring shorter documents, in retrievers like Dragon+ and Contriever. Our findings reveal significant vulnerabilities: retrievers often rely on superficial patterns like over-prioritizing document beginnings, shorter documents, repeated entities, and literal matches. Additionally, they tend to overlook whether the document contains the query's answer, lacking deep semantic understanding. Notably, when multiple biases combine, models exhibit catastrophic performance degradation, selecting the answer-containing document in less than 3% of cases over a biased document without the answer. Furthermore, we show that these biases have direct consequences for downstream applications like RAG, where retrieval-preferred documents can mislead LLMs, resulting in a 34% performance drop than not providing any documents at all.
☆ Beyond RAG: Task-Aware KV Cache Compression for Comprehensive Knowledge Reasoning
Incorporating external knowledge in large language models (LLMs) enhances their utility across diverse applications, but existing methods have trade-offs. Retrieval-Augmented Generation (RAG) fetches evidence via similarity search, but key information may fall outside top ranked results. Long-context models can process multiple documents but are computationally expensive and limited by context window size. Inspired by students condensing study material for open-book exams, we propose task-aware key-value (KV) cache compression, which compresses external knowledge in a zero- or few-shot setup. This enables LLMs to reason efficiently over a compacted representation of all relevant information. Experiments show our approach outperforms both RAG and task-agnostic compression methods. On LongBench v2, it improves accuracy by up to 7 absolute points over RAG with a 30x compression rate, while reducing inference latency from 0.43s to 0.16s. A synthetic dataset highlights that RAG performs well when sparse evidence suffices, whereas task-aware compression is superior for broad knowledge tasks.
☆ Evaluating Answer Reranking Strategies in Time-sensitive Question Answering
Despite advancements in state-of-the-art models and information retrieval techniques, current systems still struggle to handle temporal information and to correctly answer detailed questions about past events. In this paper, we investigate the impact of temporal characteristics of answers in Question Answering (QA) by exploring several simple answer selection techniques. Our findings emphasize the role of temporal features in selecting the most relevant answers from diachronic document collections and highlight differences between explicit and implicit temporal questions.
♻ ☆ Decoupled Recommender Systems: Exploring Alternative Recommender Ecosystem Designs
Recommender ecosystems are an emerging subject of research. Such research examines how the characteristics of algorithms, recommendation consumers, and item providers influence system dynamics and long-term outcomes. One architectural possibility that has not yet been widely explored in this line of research is the consequences of a configuration in which recommendation algorithms are decoupled from the platforms they serve. This is sometimes called "the friendly neighborhood algorithm store" or "middleware" model. We are particularly interested in how such architectures might offer a range of different distributions of utility across consumers, providers, and recommendation platforms. In this paper, we create a model of a recommendation ecosystem that incorporates algorithm choice and examine the outcomes of such a design.
♻ ☆ Assisting Mathematical Formalization with A Learning-based Premise Retriever
Premise selection is a crucial yet challenging step in mathematical formalization, especially for users with limited experience. Due to the lack of available formalization projects, existing approaches that leverage language models often suffer from data scarcity. In this work, we introduce an innovative method for training a premise retriever to support the formalization of mathematics. Our approach employs a BERT model to embed proof states and premises into a shared latent space. The retrieval model is trained within a contrastive learning framework and incorporates a domain-specific tokenizer along with a fine-grained similarity computation method. Experimental results show that our model is highly competitive compared to existing baselines, achieving strong performance while requiring fewer computational resources. Performance is further enhanced through the integration of a re-ranking module. To streamline the formalization process, we will release a search engine that enables users to query Mathlib theorems directly using proof states, significantly improving accessibility and efficiency. Codes are available at https://github.com/ruc-ai4math/Premise-Retrieval.
♻ ☆ SRAG: Structured Retrieval-Augmented Generation for Multi-Entity Question Answering over Wikipedia Graph
Multi-entity question answering (MEQA) poses significant challenges for large language models (LLMs), which often struggle to consolidate scattered information across multiple documents. An example question might be "What is the distribution of IEEE Fellows among various fields of study?", which requires retrieving information from diverse sources e.g., Wikipedia pages. The effectiveness of current retrieval-augmented generation (RAG) methods is limited by the LLMs' capacity to aggregate insights from numerous pages. To address this gap, this paper introduces a structured RAG (SRAG) framework that systematically organizes extracted entities into relational tables (e.g., tabulating entities with schema columns like "name" and "field of study") and then apply table-based reasoning techniques. Our approach decouples retrieval and reasoning, enabling LLMs to focus on structured data analysis rather than raw text aggregation. Extensive experiments on Wikipedia-based multi-entity QA tasks demonstrate that SRAG significantly outperforms state-of-the-art long-context LLMs and RAG solutions, achieving a 29.6% improvement in accuracy. The results underscore the efficacy of structuring unstructured data to enhance LLMs' reasoning capabilities.
♻ ☆ Evaluating Search Engines and Large Language Models for Answering Health Questions
Search engines (SEs) have traditionally been primary tools for information seeking, but the new Large Language Models (LLMs) are emerging as powerful alternatives, particularly for question-answering tasks. This study compares the performance of four popular SEs, seven LLMs, and retrieval-augmented (RAG) variants in answering 150 health-related questions from the TREC Health Misinformation (HM) Track. Results reveal SEs correctly answer between 50 and 70% of questions, often hindered by many retrieval results not responding to the health question. LLMs deliver higher accuracy, correctly answering about 80% of questions, though their performance is sensitive to input prompts. RAG methods significantly enhance smaller LLMs' effectiveness, improving accuracy by up to 30% by integrating retrieval evidence.
♻ ☆ Semi-Parametric Retrieval via Binary Bag-of-Tokens Index
Information retrieval has transitioned from standalone systems into essential components across broader applications, with indexing efficiency, cost-effectiveness, and freshness becoming increasingly critical yet often overlooked. In this paper, we introduce SemI-parametric Disentangled Retrieval (SiDR), a bi-encoder retrieval framework that decouples retrieval index from neural parameters to enable efficient, low-cost, and parameter-agnostic indexing for emerging use cases. Specifically, in addition to using embeddings as indexes like existing neural retrieval methods, SiDR supports a non-parametric tokenization index for search, achieving BM25-like indexing complexity with significantly better effectiveness. Our comprehensive evaluation across 16 retrieval benchmarks demonstrates that SiDR outperforms both neural and term-based retrieval baselines under the same indexing workload: (i) When using an embedding-based index, SiDR exceeds the performance of conventional neural retrievers while maintaining similar training complexity; (ii) When using a tokenization-based index, SiDR drastically reduces indexing cost and time, matching the complexity of traditional term-based retrieval, while consistently outperforming BM25 on all in-domain datasets; (iii) Additionally, we introduce a late parametric mechanism that matches BM25 index preparation time while outperforming other neural retrieval baselines in effectiveness.
♻ ☆ Exploiting Distribution Constraints for Scalable and Efficient Image Retrieval
Image retrieval is crucial in robotics and computer vision, with downstream applications in robot place recognition and vision-based product recommendations. Modern retrieval systems face two key challenges: scalability and efficiency. State-of-the-art image retrieval systems train specific neural networks for each dataset, an approach that lacks scalability. Furthermore, since retrieval speed is directly proportional to embedding size, existing systems that use large embeddings lack efficiency. To tackle scalability, recent works propose using off-the-shelf foundation models. However, these models, though applicable across datasets, fall short in achieving performance comparable to that of dataset-specific models. Our key observation is that, while foundation models capture necessary subtleties for effective retrieval, the underlying distribution of their embedding space can negatively impact cosine similarity searches. We introduce Autoencoders with Strong Variance Constraints (AE-SVC), which, when used for projection, significantly improves the performance of foundation models. We provide an in-depth theoretical analysis of AE-SVC. Addressing efficiency, we introduce Single-shot Similarity Space Distillation ((SS)$_2$D), a novel approach to learn embeddings with adaptive sizes that offers a better trade-off between size and performance. We conducted extensive experiments on four retrieval datasets, including Stanford Online Products (SoP) and Pittsburgh30k, using four different off-the-shelf foundation models, including DinoV2 and CLIP. AE-SVC demonstrates up to a $16\%$ improvement in retrieval performance, while (SS)$_2$D shows a further $10\%$ improvement for smaller embedding sizes.
Computation and Language 130
☆ The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
As large language models (LLMs) become more capable and agentic, the requirement for trust in their outputs grows significantly, yet at the same time concerns have been mounting that models may learn to lie in pursuit of their goals. To address these concerns, a body of work has emerged around the notion of "honesty" in LLMs, along with interventions aimed at mitigating deceptive behaviors. However, evaluations of honesty are currently highly limited, with no benchmark combining large scale and applicability to all models. Moreover, many benchmarks claiming to measure honesty in fact simply measure accuracy--the correctness of a model's beliefs--in disguise. In this work, we introduce a large-scale human-collected dataset for measuring honesty directly, allowing us to disentangle accuracy from honesty for the first time. Across a diverse set of LLMs, we find that while larger models obtain higher accuracy on our benchmark, they do not become more honest. Surprisingly, while most frontier LLMs obtain high scores on truthfulness benchmarks, we find a substantial propensity in frontier LLMs to lie when pressured to do so, resulting in low honesty scores on our benchmark. We find that simple methods, such as representation engineering interventions, can improve honesty. These results underscore the growing need for robust evaluations and effective interventions to ensure LLMs remain trustworthy.
comment: Website: https://www.mask-benchmark.ai
☆ Process-based Self-Rewarding Language Models
Large Language Models have demonstrated outstanding performance across various downstream tasks and have been widely applied in multiple scenarios. Human-annotated preference data is used for training to further improve LLMs' performance, which is constrained by the upper limit of human performance. Therefore, Self-Rewarding method has been proposed, where LLMs generate training data by rewarding their own outputs. However, the existing self-rewarding paradigm is not effective in mathematical reasoning scenarios and may even lead to a decline in performance. In this work, we propose the Process-based Self-Rewarding pipeline for language models, which introduces long-thought reasoning, step-wise LLM-as-a-Judge, and step-wise preference optimization within the self-rewarding paradigm. Our new paradigm successfully enhances the performance of LLMs on multiple mathematical reasoning benchmarks through iterative Process-based Self-Rewarding, demonstrating the immense potential of self-rewarding to achieve LLM reasoning that may surpass human capabilities.
☆ Improving LLM Safety Alignment with Dual-Objective Optimization
Existing training-time safety alignment techniques for large language models (LLMs) remain vulnerable to jailbreak attacks. Direct preference optimization (DPO), a widely deployed alignment method, exhibits limitations in both experimental and theoretical contexts as its loss function proves suboptimal for refusal learning. Through gradient-based analysis, we identify these shortcomings and propose an improved safety alignment that disentangles DPO objectives into two components: (1) robust refusal training, which encourages refusal even when partial unsafe generations are produced, and (2) targeted unlearning of harmful knowledge. This approach significantly increases LLM robustness against a wide range of jailbreak attacks, including prefilling, suffix, and multi-turn attacks across both in-distribution and out-of-distribution scenarios. Furthermore, we introduce a method to emphasize critical refusal tokens by incorporating a reward-based token-level weighting mechanism for refusal learning, which further improves the robustness against adversarial exploits. Our research also suggests that robustness to jailbreak attacks is correlated with token distribution shifts in the training process and internal representations of refusal and harmful tokens, offering valuable directions for future research in LLM safety alignment. The code is available at https://github.com/wicai24/DOOR-Alignment
☆ Effective LLM Knowledge Learning via Model Generalization
Large language models (LLMs) are trained on enormous documents that contain extensive world knowledge. However, it is still not well-understood how knowledge is acquired via autoregressive pre-training. This lack of understanding greatly hinders effective knowledge learning, especially for continued pretraining on up-to-date information, as this evolving information often lacks diverse repetitions like foundational knowledge. In this paper, we focus on understanding and improving LLM knowledge learning. We found and verified that knowledge learning for LLMs can be deemed as an implicit supervised task hidden in the autoregressive pre-training objective. Our findings suggest that knowledge learning for LLMs would benefit from methods designed to improve generalization ability for supervised tasks. Based on our analysis, we propose the formatting-based data augmentation to grow in-distribution samples, which does not present the risk of altering the facts embedded in documents as text paraphrasing. We also introduce sharpness-aware minimization as an effective optimization algorithm to better improve generalization. Moreover, our analysis and method can be readily extended to instruction tuning. Extensive experiment results validate our findings and demonstrate our methods' effectiveness in both continued pre-training and instruction tuning. This paper offers new perspectives and insights to interpret and design effective strategies for LLM knowledge learning.
☆ SoftMatcha: A Soft and Fast Pattern Matcher for Billion-Scale Corpus Searches ICLR2025
Researchers and practitioners in natural language processing and computational linguistics frequently observe and analyze the real language usage in large-scale corpora. For that purpose, they often employ off-the-shelf pattern-matching tools, such as grep, and keyword-in-context concordancers, which is widely used in corpus linguistics for gathering examples. Nonetheless, these existing techniques rely on surface-level string matching, and thus they suffer from the major limitation of not being able to handle orthographic variations and paraphrasing -- notable and common phenomena in any natural language. In addition, existing continuous approaches such as dense vector search tend to be overly coarse, often retrieving texts that are unrelated but share similar topics. Given these challenges, we propose a novel algorithm that achieves \emph{soft} (or semantic) yet efficient pattern matching by relaxing a surface-level matching with word embeddings. Our algorithm is highly scalable with respect to the size of the corpus text utilizing inverted indexes. We have prepared an efficient implementation, and we provide an accessible web tool. Our experiments demonstrate that the proposed method (i) can execute searches on billion-scale corpora in less than a second, which is comparable in speed to surface-level string matching and dense vector search; (ii) can extract harmful instances that semantically match queries from a large set of English and Japanese Wikipedia articles; and (iii) can be effectively applied to corpus-linguistic analyses of Latin, a language with highly diverse inflections.
comment: Accepted at ICLR2025
☆ Developing and Utilizing a Large-Scale Cantonese Dataset for Multi-Tasking in Large Language Models
High-quality data resources play a crucial role in learning large language models (LLMs), particularly for low-resource languages like Cantonese. Despite having more than 85 million native speakers, Cantonese is still considered a low-resource language in the field of natural language processing (NLP) due to factors such as the dominance of Mandarin, lack of cohesion within the Cantonese-speaking community, diversity in character encoding and input methods, and the tendency of overseas Cantonese speakers to prefer using English. In addition, rich colloquial vocabulary of Cantonese, English loanwords, and code-switching characteristics add to the complexity of corpus collection and processing. To address these challenges, we collect Cantonese texts from a variety of sources, including open source corpora, Hong Kong-specific forums, Wikipedia, and Common Crawl data. We conduct rigorous data processing through language filtering, quality filtering, content filtering, and de-duplication steps, successfully constructing a high-quality Cantonese corpus of over 2 billion tokens for training large language models. We further refined the model through supervised fine-tuning (SFT) on curated Cantonese tasks, enhancing its ability to handle specific applications. Upon completion of the training, the model achieves state-of-the-art (SOTA) performance on four Cantonese benchmarks. After training on our dataset, the model also exhibits improved performance on other mainstream language tasks.
☆ MAS-GPT: Training LLMs to Build LLM-based Multi-Agent Systems
LLM-based multi-agent systems (MAS) have shown significant potential in tackling diverse tasks. However, to design effective MAS, existing approaches heavily rely on manual configurations or multiple calls of advanced LLMs, resulting in inadaptability and high inference costs. In this paper, we simplify the process of building an MAS by reframing it as a generative language task, where the input is a user query and the output is a corresponding MAS. To address this novel task, we unify the representation of MAS as executable code and propose a consistency-oriented data construction pipeline to create a high-quality dataset comprising coherent and consistent query-MAS pairs. Using this dataset, we train MAS-GPT, an open-source medium-sized LLM that is capable of generating query-adaptive MAS within a single LLM inference. The generated MAS can be seamlessly applied to process user queries and deliver high-quality responses. Extensive experiments on 9 benchmarks and 5 LLMs show that the proposed MAS-GPT consistently outperforms 10+ baseline MAS methods on diverse settings, indicating MAS-GPT's high effectiveness, efficiency and strong generalization ability. Code will be available at https://github.com/rui-ye/MAS-GPT.
comment: 26 pages, 7 figures
☆ Quantification of Tenseness in English and Japanese Tense-Lax Vowels: A Lagrangian Model with Indicator $θ_1$ and Force of Tenseness Ftense(t)
The concept of vowel tenseness has traditionally been examined through the binary distinction of tense and lax vowels. However, no universally accepted quantitative definition of tenseness has been established in any language. Previous studies, including those by Jakobson, Fant, and Halle (1951) and Chomsky and Halle (1968), have explored the relationship between vowel tenseness and the vocal tract. Building on these foundations, Ishizaki (2019, 2022) proposed an indirect quantification of vowel tenseness using formant angles $\theta_1$ and $\theta_{F1}$ and their first and second derivatives, $d^Z_1(t)/dt = \lim \tan \theta_1(t$) and $d^2 Z_1(t)/dt^2 = d/dt \lim \tan \theta_1(t)$. This study extends this approach by investigating the potential role of a force-related parameter in determining vowel quality. Specifically, we introduce a simplified model based on the Lagrangian equation to describe the dynamic interaction of the tongue and jaw within the oral cavity during the articulation of close vowels. This model provides a theoretical framework for estimating the forces involved in vowel production across different languages, offering new insights into the physical mechanisms underlying vowel articulation. The findings suggest that this force-based perspective warrants further exploration as a key factor in phonetic and phonological studies.
☆ Attentive Reasoning Queries: A Systematic Method for Optimizing Instruction-Following in Large Language Models
We present Attentive Reasoning Queries (ARQs), a novel structured reasoning approach that significantly improves instruction-following in Large Language Models through domain-specialized reasoning blueprints. While LLMs demonstrate remarkable capabilities across diverse tasks, they often fail to maintain adherence to complex, use-case-specific instructions during multi-turn conversations, presenting challenges for business-critical applications. ARQs address this limitation by guiding LLMs through systematic reasoning steps with targeted queries that reinstate critical instructions and facilitate intermediate reasoning throughout the completion process. In extensive testing within Parlant, our framework for reliable customer-facing agents in which ARQs were born out of necessity, they achieved a 90.2% success rate across 87 test scenarios, outperforming both Chain-of-Thought reasoning (86.1%) and direct response generation (81.5%). ARQs showed particular strength in addressing persistent failure modes like guideline re-application and hallucination prevention. Our analysis also revealed that ARQs can potentially be more computationally efficient than free-form reasoning when carefully designed. These findings demonstrate that structured reasoning approaches provide effective mechanisms for controlling how LLMs process information and make decisions in complex scenarios.
comment: Supplementary materials, including code, is available on our GitHub: https://github.com/emcie-co/parlant/tree/arqs-a-systematic-method-for-optimizing-instruction-following-in-llms
☆ Analogical Reasoning Inside Large Language Models: Concept Vectors and the Limits of Abstraction
Analogical reasoning relies on conceptual abstractions, but it is unclear whether Large Language Models (LLMs) harbor such internal representations. We explore distilled representations from LLM activations and find that function vectors (FVs; Todd et al., 2024) - compact representations for in-context learning (ICL) tasks - are not invariant to simple input changes (e.g., open-ended vs. multiple-choice), suggesting they capture more than pure concepts. Using representational similarity analysis (RSA), we localize a small set of attention heads that encode invariant concept vectors (CVs) for verbal concepts like "antonym". These CVs function as feature detectors that operate independently of the final output - meaning that a model may form a correct internal representation yet still produce an incorrect output. Furthermore, CVs can be used to causally guide model behaviour. However, for more abstract concepts like "previous" and "next", we do not observe invariant linear representations, a finding we link to generalizability issues LLMs display within these domains.
comment: Preprint
☆ Improving Neutral Point of View Text Generation through Parameter-Efficient Reinforcement Learning and a Small-Scale High-Quality Dataset
This paper describes the construction of a dataset and the evaluation of training methods to improve generative large language models' (LLMs) ability to answer queries on sensitive topics with a Neutral Point of View (NPOV), i.e., to provide significantly more informative, diverse and impartial answers. The dataset, the SHQ-NPOV dataset, comprises 300 high-quality, human-written quadruplets: a query on a sensitive topic, an answer, an NPOV rating, and a set of links to source texts elaborating the various points of view. The first key contribution of this paper is a new methodology to create such datasets through iterative rounds of human peer-critique and annotator training, which we release alongside the dataset. The second key contribution is the identification of a highly effective training regime for parameter-efficient reinforcement learning (PE-RL) to improve NPOV generation. We compare and extensively evaluate PE-RL and multiple baselines-including LoRA finetuning (a strong baseline), SFT and RLHF. PE-RL not only improves on overall NPOV quality compared to the strongest baseline ($97.06\%\rightarrow 99.08\%$), but also scores much higher on features linguists identify as key to separating good answers from the best answers ($60.25\%\rightarrow 85.21\%$ for presence of supportive details, $68.74\%\rightarrow 91.43\%$ for absence of oversimplification). A qualitative analysis corroborates this. Finally, our evaluation finds no statistical differences between results on topics that appear in the training dataset and those on separated evaluation topics, which provides strong evidence that our approach to training PE-RL exhibits very effective out of topic generalization.
☆ Token-Level Privacy in Large Language Models
The use of language models as remote services requires transmitting private information to external providers, raising significant privacy concerns. This process not only risks exposing sensitive data to untrusted service providers but also leaves it vulnerable to interception by eavesdroppers. Existing privacy-preserving methods for natural language processing (NLP) interactions primarily rely on semantic similarity, overlooking the role of contextual information. In this work, we introduce dchi-stencil, a novel token-level privacy-preserving mechanism that integrates contextual and semantic information while ensuring strong privacy guarantees under the dchi differential privacy framework, achieving 2epsilon-dchi-privacy. By incorporating both semantic and contextual nuances, dchi-stencil achieves a robust balance between privacy and utility. We evaluate dchi-stencil using state-of-the-art language models and diverse datasets, achieving comparable and even better trade-off between utility and privacy compared to existing methods. This work highlights the potential of dchi-stencil to set a new standard for privacy-preserving NLP in modern, high-risk applications.
☆ Psy-Copilot: Visual Chain of Thought for Counseling
Large language models (LLMs) are becoming increasingly popular in the field of psychological counseling. However, when human therapists work with LLMs in therapy sessions, it is hard to understand how the model gives the answers. To address this, we have constructed Psy-COT, a graph designed to visualize the thought processes of LLMs during therapy sessions. The Psy-COT graph presents semi-structured counseling conversations alongside step-by-step annotations that capture the reasoning and insights of therapists. Moreover, we have developed Psy-Copilot, which is a conversational AI assistant designed to assist human psychological therapists in their consultations. It can offer traceable psycho-information based on retrieval, including response candidates, similar dialogue sessions, related strategies, and visual traces of results. We have also built an interactive platform for AI-assisted counseling. It has an interface that displays the relevant parts of the retrieval sub-graph. The Psy-Copilot is designed not to replace psychotherapists but to foster collaboration between AI and human therapists, thereby promoting mental health development. Our code and demo are both open-sourced and available for use.
☆ Psy-Insight: Explainable Multi-turn Bilingual Dataset for Mental Health Counseling
The in-context learning capabilities of large language models (LLMs) show great potential in mental health support. However, the lack of counseling datasets, particularly in Chinese corpora, restricts their application in this field. To address this, we constructed Psy-Insight, the first mental health-oriented explainable multi-task bilingual dataset. We collected face-to-face multi-turn counseling dialogues, which are annotated with multi-task labels and conversation process explanations. Our annotations include psychotherapy, emotion, strategy, and topic labels, as well as turn-level reasoning and session-level guidance. Psy-Insight is not only suitable for tasks such as label recognition but also meets the need for training LLMs to act as empathetic counselors through logical reasoning. Experiments show that training LLMs on Psy-Insight enables the models to not only mimic the conversation style but also understand the underlying strategies and reasoning of counseling.
☆ Feature-Level Insights into Artificial Text Detection with Sparse Autoencoders
Artificial Text Detection (ATD) is becoming increasingly important with the rise of advanced Large Language Models (LLMs). Despite numerous efforts, no single algorithm performs consistently well across different types of unseen text or guarantees effective generalization to new LLMs. Interpretability plays a crucial role in achieving this goal. In this study, we enhance ATD interpretability by using Sparse Autoencoders (SAE) to extract features from Gemma-2-2b residual stream. We identify both interpretable and efficient features, analyzing their semantics and relevance through domain- and model-specific statistics, a steering approach, and manual or LLM-based interpretation. Our methods offer valuable insights into how texts from various models differ from human-written content. We show that modern LLMs have a distinct writing style, especially in information-dense domains, even though they can produce human-like outputs with personalized prompts.
☆ Small but Mighty: Enhancing Time Series Forecasting with Lightweight LLMs
While LLMs have demonstrated remarkable potential in time series forecasting, their practical deployment remains constrained by excessive computational demands and memory footprints. Existing LLM-based approaches typically suffer from three critical limitations: Inefficient parameter utilization in handling numerical time series patterns; Modality misalignment between continuous temporal signals and discrete text embeddings; and Inflexibility for real-time expert knowledge integration. We present SMETimes, the first systematic investigation of sub-3B parameter SLMs for efficient and accurate time series forecasting. Our approach centers on three key innovations: A statistically-enhanced prompting mechanism that bridges numerical time series with textual semantics through descriptive statistical features; A adaptive fusion embedding architecture that aligns temporal patterns with language model token spaces through learnable parameters; And a dynamic mixture-of-experts framework enabled by SLMs' computational efficiency, adaptively combining base predictions with domain-specific models. Extensive evaluations across seven benchmark datasets demonstrate that our 3B-parameter SLM achieves state-of-the-art performance on five primary datasets while maintaining 3.8x faster training and 5.2x lower memory consumption compared to 7B-parameter LLM baselines. Notably, the proposed model exhibits better learning capabilities, achieving 12.3% lower MSE than conventional LLM. Ablation studies validate that our statistical prompting and cross-modal fusion modules respectively contribute 15.7% and 18.2% error reduction in long-horizon forecasting tasks. By redefining the efficiency-accuracy trade-off landscape, this work establishes SLMs as viable alternatives to resource-intensive LLMs for practical time series forecasting. Code and models are available at https://github.com/xiyan1234567/SMETimes.
comment: Work in progress
☆ English K_Quantization of LLMs Does Not Disproportionately Diminish Multilingual Performance
For consumer usage of locally deployed LLMs, the GGUF format and k_quantization are invaluable tools for maintaining the performance of the original model while reducing it to sizes deployable with consumer-grade hardware. The number of bits dedicated to each weight from the original model is reduced based on how important they are thought to be during model inference. This importance is arrived at through the application of an 'importance matrix'-a relatively small text document meant to be representative of the LLM's standard use-cases. In the vast majority of quants available online, this document is primarily written in English. It was therefore an open question whether performance on English language tasks was preserved through the sacrifice of multilingual performance and whether it can be preserved with alternate importance matrices. This article investigates these hypotheses by quantizing Llama3.3 70B on importance matrices written in three languages (English, Norwegian, and Malayalam) and evaluating them on the MixEval dataset in both English and Norwegian. All experiments related to k_quantization yielded non-significant results (In all cases p > 0.237) indicating that current quantization practices do not disproportionately harm multilingual performance.
comment: 8 pages, 6 figures
☆ PowerAttention: Exponentially Scaling of Receptive Fields for Effective Sparse Attention
Large Language Models (LLMs) face efficiency bottlenecks due to the quadratic complexity of the attention mechanism when processing long contexts. Sparse attention methods offer a promising solution, but existing approaches often suffer from incomplete effective context and/or require complex implementation of pipeline. We present a comprehensive analysis of sparse attention for autoregressive LLMs from the respective of receptive field, recognize the suboptimal nature of existing methods for expanding the receptive field, and introduce PowerAttention, a novel sparse attention design that facilitates effective and complete context extension through the theoretical analysis. PowerAttention achieves exponential receptive field growth in $d$-layer LLMs, allowing each output token to attend to $2^d$ tokens, ensuring completeness and continuity of the receptive field. Experiments demonstrate that PowerAttention outperforms existing static sparse attention methods by $5\sim 40\%$, especially on tasks demanding long-range dependencies like Passkey Retrieval and RULER, while maintaining a comparable time complexity to sliding window attention. Efficiency evaluations further highlight PowerAttention's superior speedup in both prefilling and decoding phases compared with dynamic sparse attentions and full attention ($3.0\times$ faster on 128K context), making it a highly effective and user-friendly solution for processing long sequences in LLMs.
comment: for associated code, see https://github.com/w568w/PowerAttention
☆ Scaling Crowdsourced Election Monitoring: Construction and Evaluation of Classification Models for Multilingual and Cross-Domain Classification Settings
The adoption of crowdsourced election monitoring as a complementary alternative to traditional election monitoring is on the rise. Yet, its reliance on digital response volunteers to manually process incoming election reports poses a significant scaling bottleneck. In this paper, we address the challenge of scaling crowdsourced election monitoring by advancing the task of automated classification of crowdsourced election reports to multilingual and cross-domain classification settings. We propose a two-step classification approach of first identifying informative reports and then categorising them into distinct information types. We conduct classification experiments using multilingual transformer models such as XLM-RoBERTa and multilingual embeddings such as SBERT, augmented with linguistically motivated features. Our approach achieves F1-Scores of 77\% for informativeness detection and 75\% for information type classification. We conduct cross-domain experiments, applying models trained in a source electoral domain to a new target electoral domain in zero-shot and few-shot classification settings. Our results show promising potential for model transfer across electoral domains, with F1-Scores of 59\% in zero-shot and 63\% in few-shot settings. However, our analysis also reveals a performance bias in detecting informative English reports over Swahili, likely due to imbalances in the training data, indicating a need for caution when deploying classification models in real-world election scenarios.
☆ An Aspect Extraction Framework using Different Embedding Types, Learning Models, and Dependency Structure
Aspect-based sentiment analysis has gained significant attention in recent years due to its ability to provide fine-grained insights for sentiment expressions related to specific features of entities. An important component of aspect-based sentiment analysis is aspect extraction, which involves identifying and extracting aspect terms from text. Effective aspect extraction serves as the foundation for accurate sentiment analysis at the aspect level. In this paper, we propose aspect extraction models that use different types of embeddings for words and part-of-speech tags and that combine several learning models. We also propose tree positional encoding that is based on dependency parsing output to capture better the aspect positions in sentences. In addition, a new aspect extraction dataset is built for Turkish by machine translating an English dataset in a controlled setting. The experiments conducted on two Turkish datasets showed that the proposed models mostly outperform the studies that use the same datasets, and incorporating tree positional encoding increases the performance of the models.
comment: Aspect-based Sentiment Analysis, Aspect Extraction, Natural Language Processing, Machine Learning, Deep Neural Networks, Turkish
☆ CURVALID: Geometrically-guided Adversarial Prompt Detection
Adversarial prompts capable of jailbreaking large language models (LLMs) and inducing undesirable behaviours pose a significant obstacle to their safe deployment. Current mitigation strategies rely on activating built-in defence mechanisms or fine-tuning the LLMs, but the fundamental distinctions between adversarial and benign prompts are yet to be understood. In this work, we introduce CurvaLID, a novel defense framework that efficiently detects adversarial prompts by leveraging their geometric properties. It is agnostic to the type of LLM, offering a unified detection framework across diverse adversarial prompts and LLM architectures. CurvaLID builds on the geometric analysis of text prompts to uncover their underlying differences. We theoretically extend the concept of curvature via the Whewell equation into an $n$-dimensional word embedding space, enabling us to quantify local geometric properties, including semantic shifts and curvature in the underlying manifolds. Additionally, we employ Local Intrinsic Dimensionality (LID) to capture geometric features of text prompts within adversarial subspaces. Our findings reveal that adversarial prompts differ fundamentally from benign prompts in terms of their geometric characteristics. Our results demonstrate that CurvaLID delivers superior detection and rejection of adversarial queries, paving the way for safer LLM deployment. The source code can be found at https://github.com/Cancanxxx/CurvaLID
comment: 29 Pages, 5 figues
☆ Deictic Codes, Demonstratives, and Reference: A Step Toward Solving the Grounding Problem
In this paper we address the issue of grounding for experiential concepts. Given that perceptual demonstratives are a basic form of such concepts, we examine ways of fixing the referents of such demonstratives. To avoid 'encodingism', that is, relating representations to representations, we postulate that the process of reference fixing must be bottom-up and nonconceptual, so that it can break the circle of conceptual content and touch the world. For that purpose, an appropriate causal relation between representations and the world is needed. We claim that this relation is provided by spatial and object-centered attention that leads to the formation of object files through the function of deictic acts. This entire causal process takes place at a pre-conceptual level, meeting the requirement for a solution to the grounding problem. Finally we claim that our account captures fundamental insights in Putnam's and Kripke's work on "new" reference.
☆ Enhancing Spoken Discourse Modeling in Language Models Using Gestural Cues
Research in linguistics shows that non-verbal cues, such as gestures, play a crucial role in spoken discourse. For example, speakers perform hand gestures to indicate topic shifts, helping listeners identify transitions in discourse. In this work, we investigate whether the joint modeling of gestures using human motion sequences and language can improve spoken discourse modeling in language models. To integrate gestures into language models, we first encode 3D human motion sequences into discrete gesture tokens using a VQ-VAE. These gesture token embeddings are then aligned with text embeddings through feature alignment, mapping them into the text embedding space. To evaluate the gesture-aligned language model on spoken discourse, we construct text infilling tasks targeting three key discourse cues grounded in linguistic research: discourse connectives, stance markers, and quantifiers. Results show that incorporating gestures enhances marker prediction accuracy across the three tasks, highlighting the complementary information that gestures can offer in modeling spoken discourse. We view this work as an initial step toward leveraging non-verbal cues to advance spoken language modeling in language models.
☆ Open-Source Large Language Models as Multilingual Crowdworkers: Synthesizing Open-Domain Dialogues in Several Languages With No Examples in Targets and No Machine Translation
The prevailing paradigm in the domain of Open-Domain Dialogue agents predominantly focuses on the English language, encompassing both models and datasets. Furthermore, the financial and temporal investments required for crowdsourcing such datasets for finetuning are substantial, particularly when multiple languages are involved. Fortunately, advancements in Large Language Models (LLMs) have unveiled a plethora of possibilities across diverse tasks. Specifically, instruction-tuning has enabled LLMs to execute tasks based on natural language instructions, occasionally surpassing the performance of human crowdworkers. Additionally, these models possess the capability to function in various languages within a single thread. Consequently, to generate new samples in different languages, we propose leveraging these capabilities to replicate the data collection process. We introduce a pipeline for generating Open-Domain Dialogue data in multiple Target Languages using LLMs, with demonstrations provided in a unique Source Language. By eschewing explicit Machine Translation in this approach, we enhance the adherence to language-specific nuances. We apply this methodology to the PersonaChat dataset. To enhance the openness of generated dialogues and mimic real life scenarii, we added the notion of speech events corresponding to the type of conversation the speakers are involved in and also that of common ground which represents the premises of a conversation.
☆ Visualising Policy-Reward Interplay to Inform Zeroth-Order Preference Optimisation of Large Language Models
Fine-tuning LLMs with first-order methods like back-propagation is computationally intensive. Zeroth-Order (ZO) optimisation, using function evaluations instead of gradients, reduces memory usage but suffers from slow convergence in high-dimensional models. As a result, ZO research in LLMs has mostly focused on classification, overlooking more complex generative tasks. In this paper, we introduce ZOPrO, a novel ZO algorithm designed for \textit{Preference Optimisation} in LLMs. We begin by analysing the interplay between policy and reward models during traditional (first-order) Preference Optimisation, uncovering patterns in their relative updates. Guided by these insights, we adapt Simultaneous Perturbation Stochastic Approximation (SPSA) with a targeted sampling strategy to accelerate convergence. Through experiments on summarisation, machine translation, and conversational assistants, we demonstrate that our method consistently enhances reward signals while achieving convergence times comparable to first-order methods. While it falls short of some state-of-the-art methods, our work is the first to apply Zeroth-Order methods to Preference Optimisation in LLMs, going beyond classification tasks and paving the way for a largely unexplored research direction. Code and visualisations are available at https://github.com/alessioGalatolo/VisZOPrO
comment: WIP
☆ Unified Mind Model: Reimagining Autonomous Agents in the LLM Era
Large language models (LLMs) have recently demonstrated remarkable capabilities across domains, tasks, and languages (e.g., ChatGPT and GPT-4), reviving the research of general autonomous agents with human-like cognitive abilities.Such human-level agents require semantic comprehension and instruction-following capabilities, which exactly fall into the strengths of LLMs.Although there have been several initial attempts to build human-level agents based on LLMs, the theoretical foundation remains a challenging open problem. In this paper, we propose a novel theoretical cognitive architecture, the Unified Mind Model (UMM), which offers guidance to facilitate the rapid creation of autonomous agents with human-level cognitive abilities. Specifically, our UMM starts with the global workspace theory and further leverage LLMs to enable the agent with various cognitive abilities, such as multi-modal perception, planning, reasoning, tool use, learning, memory, reflection and motivation. Building upon UMM, we then develop an agent-building engine, MindOS, which allows users to quickly create domain-/task-specific autonomous agents without any programming effort.
comment: 18 pages
☆ Taxation Perspectives from Large Language Models: A Case Study on Additional Tax Penalties
How capable are large language models (LLMs) in the domain of taxation? Although numerous studies have explored the legal domain in general, research dedicated to taxation remain scarce. Moreover, the datasets used in these studies are either simplified, failing to reflect the real-world complexities, or unavailable as open source. To address this gap, we introduce PLAT, a new benchmark designed to assess the ability of LLMs to predict the legitimacy of additional tax penalties. PLAT is constructed to evaluate LLMs' understanding of tax law, particularly in cases where resolving the issue requires more than just applying related statutes. Our experiments with six LLMs reveal that their baseline capabilities are limited, especially when dealing with conflicting issues that demand a comprehensive understanding. However, we found that enabling retrieval, self-reasoning, and discussion among multiple agents with specific role assignments, this limitation can be mitigated.
comment: 5 pages
☆ RASD: Retrieval-Augmented Speculative Decoding
Speculative decoding accelerates inference in large language models (LLMs) by generating draft tokens for target model verification. Current approaches for obtaining draft tokens rely on lightweight draft models or additional model structures to generate draft tokens and retrieve context from databases. Due to the draft model's small size and limited training data, model-based speculative decoding frequently becomes less effective in out-of-domain scenarios. Additionally, the time cost of the drafting phase results in a low upper limit on acceptance length during the verification step, limiting overall efficiency. This paper proposes RASD (Retrieval-Augmented Speculative Decoding), which adopts retrieval methods to enhance model-based speculative decoding. We introduce tree pruning and tree fusion to achieve this. Specifically, we develop a pruning method based on the draft model's probability distribution to construct the optimal retrieval tree. Second, we employ the longest prefix matching algorithm to merge the tree generated by the draft model with the retrieval tree, resulting in a unified tree for verification. Experimental results demonstrate that RASD achieves state-of-the-art inference acceleration across tasks such as DocQA, Summary, Code, and In-Domain QA. Moreover, RASD exhibits strong scalability, seamlessly integrating with various speculative decoding approaches, including both generation-based and retrieval-based methods.
☆ When Claims Evolve: Evaluating and Enhancing the Robustness of Embedding Models Against Misinformation Edits
Online misinformation remains a critical challenge, and fact-checkers increasingly rely on embedding-based methods to retrieve relevant fact-checks. Yet, when debunked claims reappear in edited forms, the performance of these methods is unclear. In this work, we introduce a taxonomy of six common real-world misinformation edits and propose a perturbation framework that generates valid, natural claim variations. Our multi-stage retrieval evaluation reveals that standard embedding models struggle with user-introduced edits, while LLM-distilled embeddings offer improved robustness at a higher computational cost. Although a strong reranker helps mitigate some issues, it cannot fully compensate for first-stage retrieval gaps. Addressing these retrieval gaps, our train- and inference-time mitigation approaches enhance in-domain robustness by up to 17 percentage points and boost out-of-domain generalization by 10 percentage points over baseline models. Overall, our findings provide practical improvements to claim-matching systems, enabling more reliable fact-checking of evolving misinformation.
☆ The Serendipity of Claude AI: Case of the 13 Low-Resource National Languages of Mali
Recent advances in artificial intelligence (AI) and natural language processing (NLP) have improved the representation of underrepresented languages. However, most languages, including Mali's 13 official national languages, continue to be poorly supported or unsupported by automatic translation and generative AI. This situation appears to have slightly improved with certain recent LLM releases. The study evaluated Claude AI's translation performance on each of the 13 national languages of Mali. In addition to ChrF2 and BLEU scores, human evaluators assessed translation accuracy, contextual consistency, robustness to dialect variations, management of linguistic bias, adaptation to a limited corpus, and ease of understanding. The study found that Claude AI performs robustly for languages with very modest language resources and, while unable to produce understandable and coherent texts for Malian languages with minimal resources, still manages to produce results which demonstrate the ability to mimic some elements of the language.
Transformers for molecular property prediction: Domain adaptation efficiently improves performance
Most of the current transformer-based chemical language models are pre-trained on millions to billions of molecules. However, the improvement from such scaling in dataset size is not confidently linked to improved molecular property prediction. The aim of this study is to investigate and overcome some of the limitations of transformer models in predicting molecular properties. Specifically, we examine the impact of pre-training dataset size and diversity on the performance of transformer models and investigate the use of domain adaptation as a technique for improving model performance. First, our findings indicate that increasing pretraining dataset size beyond 400K molecules from the GuacaMol dataset does not result in a significant improvement on four ADME endpoints, namely, solubility, permeability, microsomal stability, and plasma protein binding. Second, our results demonstrate that using domain adaptation by further training the transformer model on a small set of domain-relevant molecules, i.e., a few hundred to a few thousand, using multi-task regression of physicochemical properties was sufficient to significantly improve performance for three out of the four investigated ADME endpoints (P-value < 0.001). Finally, we observe that a model pre-trained on 400K molecules and domain adopted on a few hundred/thousand molecules performs similarly (P-value > 0.05) to more complicated transformer models like MolBERT(pre-trained on 1.3M molecules) and MolFormer (pre-trained on 100M molecules). A comparison to a random forest model trained on basic physicochemical properties showed similar performance to the examined transformer models. We believe that current transformer models can be improved through further systematic analysis of pre-training and downstream data, pre-training objectives, and scaling laws, ultimately leading to better and more helpful models.
☆ EnigmaToM: Improve LLMs' Theory-of-Mind Reasoning Capabilities with Neural Knowledge Base of Entity States
Theory-of-Mind (ToM), the ability to infer others' perceptions and mental states, is fundamental to human interaction but remains a challenging task for Large Language Models (LLMs). While existing ToM reasoning methods show promise with reasoning via perceptual perspective-taking, they often rely excessively on LLMs, reducing their efficiency and limiting their applicability to high-order ToM reasoning, which requires multi-hop reasoning about characters' beliefs. To address these issues, we present EnigmaToM, a novel neuro-symbolic framework that enhances ToM reasoning by integrating a Neural Knowledge Base of entity states (Enigma) for (1) a psychology-inspired iterative masking mechanism that facilitates accurate perspective-taking and (2) knowledge injection that elicits key entity information. Enigma generates structured representations of entity states, which construct spatial scene graphs -- leveraging spatial information as an inductive bias -- for belief tracking of various ToM orders and enhancing events with fine-grained entity state details. Experimental results on multiple benchmarks, including ToMi, HiToM, and FANToM, show that EnigmaToM significantly improves ToM reasoning across LLMs of varying sizes, particularly excelling in high-order reasoning scenarios.
☆ iNews: A Multimodal Dataset for Modeling Personalized Affective Responses to News
Current approaches to emotion detection often overlook the inherent subjectivity of affective experiences, instead relying on aggregated labels that mask individual variations in emotional responses. We introduce iNews, a novel large-scale dataset explicitly capturing subjective affective responses to news headlines. Our dataset comprises annotations from 291 demographically diverse UK participants across 2,899 multimodal Facebook news posts from major UK outlets, with an average of 5.18 annotators per sample. For each post, annotators provide multifaceted labels including valence, arousal, dominance, discrete emotions, content relevance judgments, sharing likelihood, and modality importance ratings (text, image, or both). Furthermore, we collect comprehensive annotator persona information covering demographics, personality, media trust, and consumption patterns, which explain 15.2% of annotation variance - higher than existing NLP datasets. Incorporating this information yields a 7% accuracy gain in zero-shot prediction and remains beneficial even with 32-shot. iNews will enhance research in LLM personalization, subjectivity, affective computing, and individual-level behavior simulation.
LLM as GNN: Graph Vocabulary Learning for Text-Attributed Graph Foundation Models
Text-Attributed Graphs (TAGs), where each node is associated with text descriptions, are ubiquitous in real-world scenarios. They typically exhibit distinctive structure and domain-specific knowledge, motivating the development of a Graph Foundation Model (GFM) that generalizes across diverse graphs and tasks. Despite large efforts to integrate Large Language Models (LLMs) and Graph Neural Networks (GNNs) for TAGs, existing approaches suffer from decoupled architectures with two-stage alignment, limiting their synergistic potential. Even worse, existing methods assign out-of-vocabulary (OOV) tokens to graph nodes, leading to graph-specific semantics, token explosion, and incompatibility with task-oriented prompt templates, which hinders cross-graph and cross-task transferability. To address these challenges, we propose PromptGFM, a versatile GFM for TAGs grounded in graph vocabulary learning. PromptGFM comprises two key components: (1) Graph Understanding Module, which explicitly prompts LLMs to replicate the finest GNN workflow within the text space, facilitating seamless GNN-LLM integration and elegant graph-text alignment; (2) Graph Inference Module, which establishes a language-based graph vocabulary ensuring expressiveness, transferability, and scalability, enabling readable instructions for LLM fine-tuning. Extensive experiments demonstrate our superiority and transferability across diverse graphs and tasks. The code is available at this: https://github.com/agiresearch/PromptGFM.
☆ The Box is in the Pen: Evaluating Commonsense Reasoning in Neural Machine Translation EMNLP
Does neural machine translation yield translations that are congenial with common sense? In this paper, we present a test suite to evaluate the commonsense reasoning capability of neural machine translation. The test suite consists of three test sets, covering lexical and contextless/contextual syntactic ambiguity that requires commonsense knowledge to resolve. We manually create 1,200 triples, each of which contain a source sentence and two contrastive translations, involving 7 different common sense types. Language models pretrained on large-scale corpora, such as BERT, GPT-2, achieve a commonsense reasoning accuracy of lower than 72% on target translations of this test suite. We conduct extensive experiments on the test suite to evaluate commonsense reasoning in neural machine translation and investigate factors that have impact on this capability. Our experiments and analyses demonstrate that neural machine translation performs poorly on commonsense reasoning of the three ambiguity types in terms of both reasoning accuracy (60.1%) and reasoning consistency (31%). The built commonsense test suite is available at https://github.com/tjunlp-lab/CommonMT.
comment: EMNLP findings 2020
☆ SEOE: A Scalable and Reliable Semantic Evaluation Framework for Open Domain Event Detection
Automatic evaluation for Open Domain Event Detection (ODED) is a highly challenging task, because ODED is characterized by a vast diversity of un-constrained output labels from various domains. Nearly all existing evaluation methods for ODED usually first construct evaluation benchmarks with limited labels and domain coverage, and then evaluate ODED methods using metrics based on token-level label matching rules. However, this kind of evaluation framework faces two issues: (1) The limited evaluation benchmarks lack representatives of the real world, making it difficult to accurately reflect the performance of various ODED methods in real-world scenarios; (2) Evaluation metrics based on token-level matching rules fail to capture semantic similarity between predictions and golden labels. To address these two problems above, we propose a scalable and reliable Semantic-level Evaluation framework for Open domain Event detection (SEOE) by constructing a more representative evaluation benchmark and introducing a semantic evaluation metric. Specifically, our proposed framework first constructs a scalable evaluation benchmark that currently includes 564 event types covering 7 major domains, with a cost-effective supplementary annotation strategy to ensure the benchmark's representativeness. The strategy also allows for the supplement of new event types and domains in the future. Then, the proposed SEOE leverages large language models (LLMs) as automatic evaluation agents to compute a semantic F1-score, incorporating fine-grained definitions of semantically similar labels to enhance the reliability of the evaluation. Extensive experiments validate the representatives of the benchmark and the reliability of the semantic evaluation metric. Existing ODED methods are thoroughly evaluated, and the error patterns of predictions are analyzed, revealing several insightful findings.
☆ Which books do I like?
Finding enjoyable fiction books can be challenging, partly because stories are multi-faceted and one's own literary taste might be difficult to ascertain. Here, we introduce the ISAAC method (Introspection-Support, AI-Annotation, and Curation), a pipeline which supports fiction readers in gaining awareness of their literary preferences and finding enjoyable books. ISAAC consists of four steps: a user supplies book ratings, an AI agent researches and annotates the provided books, patterns in book enjoyment are reviewed by the user, and the AI agent recommends new books. In this proof-of-concept self-study, the authors test whether ISAAC can highlight idiosyncratic patterns in their book enjoyment, spark a deeper reflection about their literary tastes, and make accurate, personalized recommendations of enjoyable books and underexplored literary niches. Results highlight substantial advantages of ISAAC over existing methods such as an integration of automation and intuition, accurate and customizable annotations, and explainable book recommendations. Observed disadvantages are that ISAAC's outputs can elicit false self-narratives (if statistical patterns are taken at face value), that books cannot be annotated if their online documentation is lacking, and that people who are new to reading have to rely on assumed book ratings or movie ratings to power the ISAAC pipeline. We discuss additional opportunities of ISAAC-style book annotations for the study of literary trends, and the scientific classification of books and readers.
☆ Enhancing Abnormality Grounding for Vision Language Models with Knowledge Descriptions
Visual Language Models (VLMs) have demonstrated impressive capabilities in visual grounding tasks. However, their effectiveness in the medical domain, particularly for abnormality detection and localization within medical images, remains underexplored. A major challenge is the complex and abstract nature of medical terminology, which makes it difficult to directly associate pathological anomaly terms with their corresponding visual features. In this work, we introduce a novel approach to enhance VLM performance in medical abnormality detection and localization by leveraging decomposed medical knowledge. Instead of directly prompting models to recognize specific abnormalities, we focus on breaking down medical concepts into fundamental attributes and common visual patterns. This strategy promotes a stronger alignment between textual descriptions and visual features, improving both the recognition and localization of abnormalities in medical images.We evaluate our method on the 0.23B Florence-2 base model and demonstrate that it achieves comparable performance in abnormality grounding to significantly larger 7B LLaVA-based medical VLMs, despite being trained on only 1.5% of the data used for such models. Experimental results also demonstrate the effectiveness of our approach in both known and previously unseen abnormalities, suggesting its strong generalization capabilities.
comment: 11 pages, 3 figures
☆ LexGenie: Automated Generation of Structured Reports for European Court of Human Rights Case Law
Analyzing large volumes of case law to uncover evolving legal principles, across multiple cases, on a given topic is a demanding task for legal professionals. Structured topical reports provide an effective solution by summarizing key issues, principles, and judgments, enabling comprehensive legal analysis on a particular topic. While prior works have advanced query-based individual case summarization, none have extended to automatically generating multi-case structured reports. To address this, we introduce LexGenie, an automated LLM-based pipeline designed to create structured reports using the entire body of case law on user-specified topics within the European Court of Human Rights jurisdiction. LexGenie retrieves, clusters, and organizes relevant passages by topic to generate a structured outline and cohesive content for each section. Expert evaluation confirms LexGenie's utility in producing structured reports that enhance efficient, scalable legal analysis.
☆ Can Frontier LLMs Replace Annotators in Biomedical Text Mining? Analyzing Challenges and Exploring Solutions
Large language models (LLMs) can perform various natural language processing (NLP) tasks through in-context learning without relying on supervised data. However, multiple previous studies have reported suboptimal performance of LLMs in biological text mining. By analyzing failure patterns in these evaluations, we identified three primary challenges for LLMs in biomedical corpora: (1) LLMs fail to learn implicit dataset-specific nuances from supervised data, (2) The common formatting requirements of discriminative tasks limit the reasoning capabilities of LLMs particularly for LLMs that lack test-time compute, and (3) LLMs struggle to adhere to annotation guidelines and match exact schemas, which hinders their ability to understand detailed annotation requirements which is essential in biomedical annotation workflow. To address these challenges, we experimented with prompt engineering techniques targeted to the above issues, and developed a pipeline that dynamically extracts instructions from annotation guidelines. Our findings show that frontier LLMs can approach or surpass the performance of state-of-the-art (SOTA) BERT-based models with minimal reliance on manually annotated data and without fine-tuning. Furthermore, we performed model distillation on a closed-source LLM, demonstrating that a BERT model trained exclusively on synthetic data annotated by LLMs can also achieve a practical performance. Based on these results, we explored the feasibility of partially replacing manual annotation with LLMs in production scenarios for biomedical text mining.
☆ FANS -- Formal Answer Selection for Natural Language Math Reasoning Using Lean4
Large Language Models (LLMs) have displayed astonishing abilities in various tasks, especially in text generation, classification, question answering, etc. However, the reasoning ability of LLMs still faces many debates. The inherent ambiguity of Natural Language (NL) limits LLMs' ability to perform verifiable reasoning, making its answers lack coherence and trustworthy support. To tackle the above problems, we propose a novel framework named FANS: Formal ANswer Selection for Natural Language Math Reasoning Using Lean4. To the best of our knowledge, it is the first framework that utilizes Lean4 to enhance LLMs' NL math reasoning ability. In particular, given an NL math question and LLM-generated answers, FANS first translates it into Lean4 theorem statements. Then it tries to prove it using a Lean4 prover and verify it by Lean4. Finally, it uses the FL result to assist in answer selection. It enhances LLMs' NL math ability in providing a computer-verifiable solution for its correct answer and proposes an alternative method for answer selection beyond the reward model. Extensive experiments indicate the effectiveness of our framework. It can improve the accuracy rate of reward model enhanced LLMs in the MATH-500 dataset by at most 1.91% and AMC-23 by at most 8.33% on strong reward-model baselines. In some particular fields like number theory that Lean4 experts in, we can even select all correct solutions. The qualitative analysis also shows our framework can make NL results formally backed by Lean4 proofs. As a pioneering work in the corresponding field, we will open-source all our models and datasets to further boost the development of the field.
☆ Targeted Distillation for Sentiment Analysis
This paper presents a compact model that achieves strong sentiment analysis capabilities through targeted distillation from advanced large language models (LLMs). Our methodology decouples the distillation target into two key components: sentiment-related knowledge and task alignment. To transfer these components, we propose a two-stage distillation framework. The first stage, knowledge-driven distillation (\textsc{KnowDist}), transfers sentiment-related knowledge to enhance fundamental sentiment analysis capabilities. The second stage, in-context learning distillation (\textsc{ICLDist}), transfers task-specific prompt-following abilities to optimize task alignment. For evaluation, we introduce \textsc{SentiBench}, a comprehensive sentiment analysis benchmark comprising 3 task categories across 12 datasets. Experiments on this benchmark demonstrate that our model effectively balances model size and performance, showing strong competitiveness compared to existing small-scale LLMs.
☆ MA-LoT: Multi-Agent Lean-based Long Chain-of-Thought Reasoning enhances Formal Theorem Proving
Solving mathematical problems using computer-verifiable languages like Lean has significantly impacted mathematical and computer science communities. State-of-the-art methods utilize single Large Language Models (LLMs) as agents or provers to either generate complete proof or perform tree searches. However, single-agent methods inherently lack a structured way to combine high-level reasoning in Natural Language (NL) with Formal Language (FL) verification feedback. To solve these issues, we propose MA-LoT: Multi-Agent Lean-based Long Chain-of-Thought framework, (to the best of our knowledge), the first multi-agent framework for Lean4 theorem proving that balance high-level NL reasoning and FL verification in Long CoT. Using this structured interaction, our approach enables deeper insights and long-term coherence in proof generation, with which past methods struggle. We do this by leveraging emergent formal reasoning ability in Long CoT using our novel LoT-Transfer Learning training-inference pipeline. Extensive experiments show that our framework achieves 54.51% accuracy rate on the Lean4 version of MiniF2F-Test dataset, largely outperforming GPT-4 (22.95%), single-agent tree search (InternLM-Step-Prover, 50.70%), and whole-proof generation (DeepSeek-Prover-v1.5, 48.36%) baselines. Furthermore, our findings highlight the potential of combining Long CoT with formal verification for a more insightful generation in a broader perspective.
☆ Towards Robust Universal Information Extraction: Benchmark, Evaluation, and Solution
In this paper, we aim to enhance the robustness of Universal Information Extraction (UIE) by introducing a new benchmark dataset, a comprehensive evaluation, and a feasible solution. Existing robust benchmark datasets have two key limitations: 1) They generate only a limited range of perturbations for a single Information Extraction (IE) task, which fails to evaluate the robustness of UIE models effectively; 2) They rely on small models or handcrafted rules to generate perturbations, often resulting in unnatural adversarial examples. Considering the powerful generation capabilities of Large Language Models (LLMs), we introduce a new benchmark dataset for Robust UIE, called RUIE-Bench, which utilizes LLMs to generate more diverse and realistic perturbations across different IE tasks. Based on this dataset, we comprehensively evaluate existing UIE models and reveal that both LLM-based models and other models suffer from significant performance drops. To improve robustness and reduce training costs, we propose a data-augmentation solution that dynamically selects hard samples for iterative training based on the model's inference loss. Experimental results show that training with only \textbf{15\%} of the data leads to an average \textbf{7.5\%} relative performance improvement across three IE tasks.
☆ Structured Outputs Enable General-Purpose LLMs to be Medical Experts
Medical question-answering (QA) is a critical task for evaluating how effectively large language models (LLMs) encode clinical knowledge and assessing their potential applications in medicine. Despite showing promise on multiple-choice tests, LLMs frequently struggle with open-ended medical questions, producing responses with dangerous hallucinations or lacking comprehensive coverage of critical aspects. Existing approaches attempt to address these challenges through domain-specific fine-tuning, but this proves resource-intensive and difficult to scale across models. To improve the comprehensiveness and factuality of medical responses, we propose a novel approach utilizing structured medical reasoning. Our method guides LLMs through an seven-step cognitive process inspired by clinical diagnosis, enabling more accurate and complete answers without additional training. Experiments on the MedLFQA benchmark demonstrate that our approach achieves the highest Factuality Score of 85.8, surpassing fine-tuned models. Notably, this improvement transfers to smaller models, highlighting the method's efficiency and scalability. Our code and datasets are available.
☆ Designing Speech Technologies for Australian Aboriginal English: Opportunities, Risks and Participation
In Australia, post-contact language varieties, including creoles and local varieties of international languages, emerged as a result of forced contact between Indigenous communities and English speakers. These contact varieties are widely used, yet are poorly supported by language technologies. This gap presents barriers to participation in civil and economic society for Indigenous communities using these varieties, and reproduces minoritisation of contemporary Indigenous sociolinguistic identities. This paper concerns three questions regarding this context. First, can speech technologies support speakers of Australian Aboriginal English, a local indigenised variety of English? Second, what risks are inherent in such a project? Third, what technology development practices are appropriate for this context, and how can researchers integrate meaningful community participation in order to mitigate risks? We argue that opportunities do exist -- as well as risks -- and demonstrate this through a case study exploring design practices in a real-world project aiming to improve speech technologies for Australian Aboriginal English. We discuss how we integrated culturally appropriate and participatory processes throughout the project. We call for increased support for languages used by Indigenous communities, including contact varieties, which provide practical economic and socio-cultural benefits, provided that participatory and culturally safe practices are enacted.
☆ Intermediate-Task Transfer Learning: Leveraging Sarcasm Detection for Stance Detection
Stance Detection (SD) on social media has emerged as a prominent area of interest with implications for social business and political applications thereby garnering escalating research attention within NLP. The inherent subtlety and complexity of texts procured from online platforms pose challenges for SD algorithms in accurately discerning the authors stance. Mostly the inclusion of sarcastic and figurative language drastically impacts the performance of SD models. This paper addresses this by employing sarcasm detection intermediate-task transfer learning tailored for SD. The proposed methodology involves the finetuning of BERT and RoBERTa and the concatenation of convolutional BiLSTM and dense layers. Rigorous experiments are conducted on publicly available datasets to evaluate our transfer-learning framework. The performance of the approach is assessed against various State-Of-The-Art baselines for SD providing empirical evidence of its effectiveness. Notably our model outperforms the best SOTA models even prior to sarcasm-detection pretraining. The integration of sarcasm knowledge into the model proves instrumental in mitigating misclassifications of sarcastic textual elements in SD. Our model accurately predicts 85% of texts that were previously misclassified by the model without sarcasm-detection pretraining thereby amplifying the average F1-score of the model. Our experiments also revealed that the success of the transfer-learning framework is contingent upon the correlation of lexical attributes between the intermediate task and the target task. This study represents the first exploration of sarcasm detection as an intermediate transfer-learning task in the context of SD and simultaneously uses the concatenation of BERT or RoBERTa with other deep-learning techniques establishing the proposed approach as a foundational baseline for future research endeavors in this domain.
comment: 8 pages, 2 figures, published in The Sixteenth International Conference on Information (eKNOW 2024)
☆ DSVD: Dynamic Self-Verify Decoding for Faithful Generation in Large Language Models
The reliability of large language models remains a critical challenge, particularly due to their susceptibility to hallucinations and factual inaccuracies during text generation. Existing solutions either underutilize models' self-correction with preemptive strategies or use costly post-hoc verification. To further explore the potential of real-time self-verification and correction, we present Dynamic Self-Verify Decoding (DSVD), a novel decoding framework that enhances generation reliability through real-time hallucination detection and efficient error correction. DSVD integrates two key components: (1) parallel self-verification architecture for continuous quality assessment, (2) dynamic rollback mechanism for targeted error recovery. Extensive experiments across five benchmarks demonstrate DSVD's effectiveness, achieving significant improvement in truthfulness (Quesetion-Answering) and factual accuracy (FActScore). Results show the DSVD can be further incorporated with existing faithful decoding methods to achieve stronger performance. Our work establishes that real-time self-verification during generation offers a viable path toward more trustworthy language models without sacrificing practical deployability.
☆ Towards Understanding Multi-Round Large Language Model Reasoning: Approximability, Learnability and Generalizability
Recent advancements in cognitive science and multi-round reasoning techniques for Large Language Models (LLMs) suggest that iterative thinking processes improve problem-solving performance in complex tasks. Inspired by this, approaches like Chain-of-Thought, debating, and self-refinement have been applied to auto-regressive LLMs, achieving significant successes in tasks such as mathematical reasoning, commonsense reasoning, and multi-hop question answering. Despite these successes, the theoretical basis for how multi-round reasoning enhances problem-solving abilities remains underexplored. In this work, we investigate the approximation, learnability, and generalization properties of multi-round auto-regressive models. We show that Transformers with finite context windows are universal approximators for steps of Turing-computable functions and can approximate any Turing-computable sequence-to-sequence function through multi-round reasoning. We extend PAC learning to sequence generation and demonstrate that multi-round generation is learnable even when the sequence length exceeds the model's context window. Finally, we examine how generalization error propagates across rounds, and show how the aforementioned approaches can help constrain this error, ensuring outputs stay within an expectation boundary. This work sheds light on the systemic theoretical foundations of multi-round sequence learning and reasoning, emphasizing its role in inference complexity.
☆ The Devil Is in the Details: Tackling Unimodal Spurious Correlations for Generalizable Multimodal Reward Models
Multimodal Reward Models (MM-RMs) are crucial for aligning Large Language Models (LLMs) with human preferences, particularly as LLMs increasingly interact with multimodal data. However, we find that MM-RMs trained on existing datasets often struggle to generalize to out-of-distribution data due to their reliance on unimodal spurious correlations, primarily text-only shortcuts within the training distribution, which prevents them from leveraging true multimodal reward functions. To address this, we introduce a Shortcut-aware MM-RM learning algorithm that mitigates this issue by dynamically reweighting training samples, shifting the distribution toward better multimodal understanding, and reducing dependence on unimodal spurious correlations. Our experiments demonstrate significant improvements in generalization, downstream task performance, and scalability, establishing a more robust framework for multimodal reward modeling.
☆ External Reliable Information-enhanced Multimodal Contrastive Learning for Fake News Detection AAAI'25
With the rapid development of the Internet, the information dissemination paradigm has changed and the efficiency has been improved greatly. While this also brings the quick spread of fake news and leads to negative impacts on cyberspace. Currently, the information presentation formats have evolved gradually, with the news formats shifting from texts to multimodal contents. As a result, detecting multimodal fake news has become one of the research hotspots. However, multimodal fake news detection research field still faces two main challenges: the inability to fully and effectively utilize multimodal information for detection, and the low credibility or static nature of the introduced external information, which limits dynamic updates. To bridge the gaps, we propose ERIC-FND, an external reliable information-enhanced multimodal contrastive learning framework for fake news detection. ERIC-FND strengthens the representation of news contents by entity-enriched external information enhancement method. It also enriches the multimodal news information via multimodal semantic interaction method where the multimodal constrative learning is employed to make different modality representations learn from each other. Moreover, an adaptive fusion method is taken to integrate the news representations from different dimensions for the eventual classification. Experiments are done on two commonly used datasets in different languages, X (Twitter) and Weibo. Experiment results demonstrate that our proposed model ERIC-FND outperforms existing state-of-the-art fake news detection methods under the same settings.
comment: accepted by AAAI'25
☆ Monitoring Decoding: Mitigating Hallucination via Evaluating the Factuality of Partial Response during Generation
While large language models have demonstrated exceptional performance across a wide range of tasks, they remain susceptible to hallucinations -- generating plausible yet factually incorrect contents. Existing methods to mitigating such risk often rely on sampling multiple full-length generations, which introduces significant response latency and becomes ineffective when the model consistently produces hallucinated outputs with high confidence. To address these limitations, we introduce Monitoring Decoding (MD), a novel framework that dynamically monitors the generation process and selectively applies in-process interventions, focusing on revising crucial tokens responsible for hallucinations. Instead of waiting until completion of multiple full-length generations, we identify hallucination-prone tokens during generation using a monitor function, and further refine these tokens through a tree-based decoding strategy. This approach ensures an enhanced factual accuracy and coherence in the generated output while maintaining efficiency. Experimental results demonstrate that MD consistently outperforms self-consistency-based approaches in both effectiveness and efficiency, achieving higher factual accuracy while significantly reducing computational overhead.
☆ MuCo-KGC: Multi-Context-Aware Knowledge Graph Completion
Knowledge graph completion (KGC) seeks to predict missing entities (e.g., heads or tails) or relationships in knowledge graphs (KGs), which often contain incomplete data. Traditional embedding-based methods, such as TransE and ComplEx, have improved tail entity prediction but struggle to generalize to unseen entities during testing. Textual-based models mitigate this issue by leveraging additional semantic context; however, their reliance on negative triplet sampling introduces high computational overhead, semantic inconsistencies, and data imbalance. Recent approaches, like KG-BERT, show promise but depend heavily on entity descriptions, which are often unavailable in KGs. Critically, existing methods overlook valuable structural information in the KG related to the entities and relationships. To address these challenges, we propose Multi-Context-Aware Knowledge Graph Completion (MuCo-KGC), a novel model that utilizes contextual information from linked entities and relations within the graph to predict tail entities. MuCo-KGC eliminates the need for entity descriptions and negative triplet sampling, significantly reducing computational complexity while enhancing performance. Our experiments on standard datasets, including FB15k-237, WN18RR, CoDEx-S, and CoDEx-M, demonstrate that MuCo-KGC outperforms state-of-the-art methods on three datasets. Notably, MuCo-KGC improves MRR on WN18RR, and CoDEx-S and CoDEx-M datasets by $1.63\%$, and $3.77\%$ and $20.15\%$ respectively, demonstrating its effectiveness for KGC tasks.
☆ Preliminary Report: Enhancing Role Differentiation in Conversational HCI Through Chromostereopsis
We propose leveraging chromostereopsis, a perceptual phenomenon inducing depth perception through color contrast, as a novel approach to visually differentiating conversational roles in text-based AI interfaces. This method aims to implicitly communicate role hierarchy and add a subtle sense of physical space.
comment: Preliminary Report, 8 pages, 1 figures
☆ On the Acquisition of Shared Grammatical Representations in Bilingual Language Models
While crosslingual transfer is crucial to contemporary language models' multilingual capabilities, how it occurs is not well understood. In this paper, we ask what happens to a monolingual language model when it begins to be trained on a second language. Specifically, we train small bilingual models for which we control the amount of data for each language and the order of language exposure. To find evidence of shared multilingual representations, we turn to structural priming, a method used to study grammatical representations in humans. We first replicate previous crosslingual structural priming results and find that after controlling for training data quantity and language exposure, there are asymmetrical effects across language pairs and directions. We argue that this asymmetry may shape hypotheses about human structural priming effects. We also find that structural priming effects are less robust for less similar language pairs, highlighting potential limitations of crosslingual transfer learning and shared representations for typologically diverse languages.
☆ Performance Comparison of Large Language Models on Advanced Calculus Problems
This paper presents an in-depth analysis of the performance of seven different Large Language Models (LLMs) in solving a diverse set of math advanced calculus problems. The study aims to evaluate these models' accuracy, reliability, and problem-solving capabilities, including ChatGPT 4o, Gemini Advanced with 1.5 Pro, Copilot Pro, Claude 3.5 Sonnet, Meta AI, Mistral AI, and Perplexity. The assessment was conducted through a series of thirty-two test problems, encompassing a total of 320 points. The problems covered various topics, from vector calculations and geometric interpretations to integral evaluations and optimization tasks. The results highlight significant trends and patterns in the models' performance, revealing both their strengths and weaknesses - for instance, models like ChatGPT 4o and Mistral AI demonstrated consistent accuracy across various problem types, indicating their robustness and reliability in mathematical problem-solving, while models such as Gemini Advanced with 1.5 Pro and Meta AI exhibited specific weaknesses, particularly in complex problems involving integrals and optimization, suggesting areas for targeted improvements. The study also underscores the importance of re-prompting in achieving accurate solutions, as seen in several instances where models initially provided incorrect answers but corrected them upon re-prompting. Overall, this research provides valuable insights into the current capabilities and limitations of LLMs in the domain of math calculus, with the detailed analysis of each model's performance on specific problems offering a comprehensive understanding of their strengths and areas for improvement, contributing to the ongoing development and refinement of LLM technology. The findings are particularly relevant for educators, researchers, and developers seeking to leverage LLMs for educational and practical applications in mathematics.
☆ Tec-Habilidad: Skill Classification for Bridging Education and Employment
Job application and assessment processes have evolved significantly in recent years, largely due to advancements in technology and changes in the way companies operate. Skill extraction and classification remain an important component of the modern hiring process as it provides a more objective way to evaluate candidates and automatically align their skills with the job requirements. However, to effectively evaluate the skills, the skill extraction tools must recognize varied mentions of skills on resumes, including direct mentions, implications, synonyms, acronyms, phrases, and proficiency levels, and differentiate between hard and soft skills. While tools like LLMs (Large Model Models) help extract and categorize skills from job applications, there's a lack of comprehensive datasets for evaluating the effectiveness of these models in accurately identifying and classifying skills in Spanish-language job applications. This gap hinders our ability to assess the reliability and precision of the models, which is crucial for ensuring that the selected candidates truly possess the required skills for the job. In this paper, we develop a Spanish language dataset for skill extraction and classification, provide annotation methodology to distinguish between knowledge, skill, and abilities, and provide deep learning baselines to advance robust solutions for skill classification.
☆ Personalized Federated Fine-tuning for Heterogeneous Data: An Automatic Rank Learning Approach via Two-Level LoRA
We study the task of personalized federated fine-tuning with heterogeneous data in the context of language models, where clients collaboratively fine-tune a language model (e.g., BERT, GPT) without sharing their local data, achieving personalization simultaneously. While recent efforts have applied parameter-efficient fine-tuning techniques like low-rank adaptation (LoRA) in federated settings, they typically use single or multiple independent low-rank adapters with predefined maximal and minimal ranks, which may not be optimal for diverse data sources over clients. To address this issue, we propose PF2LoRA, a new personalized federated fine-tuning algorithm built on a novel \emph{automatic rank learning approach via two-level LoRA}. Given the pretrained language model whose weight is frozen, our algorithm aims to learn two levels of adaptation simultaneously: the first level aims to learn a common adapter for all clients, while the second level fosters individual client personalization. A key advantage of PF2LoRA is its ability to adaptively determine a suitable rank based on an individual client's data, rather than relying on a predefined rank that is agnostic to data heterogeneity. We present a synthetic example that highlights how PF2LoRA automatically learns the ground-truth rank for each client, tailoring the adaptation to match the properties of their individual data. Notably, this approach introduces minimal additional memory overhead, as the second-level adaptation comprises a small number of parameters compared to the first level. Our experiments on natural language understanding and generation tasks demonstrate that PF2LoRA significantly outperforms existing federated fine-tuning methods.
comment: 28 pages, 5 figures
☆ LEWIS (LayEr WIse Sparsity) -- A Training Free Guided Model Merging Approach ICLR 2025
As specialized large language models (LLMs) become increasingly prevalent, model merging methods are being used to combine them to create a single multi-task model without requiring any additional data or training. However, these approaches fall short when the objective of merging is to increase the downstream model's performance on a particular task-specific benchmark. In this work, we propose LEWIS (Layer Wise Sparsity), a guided model-merging framework that uses activation-based layer importance to dynamically adjust layer-wise task-vector sparsity required for the merge process. LEWIS uses a calibration dataset to prioritize critical layers during the task-vector pruning process required for model merging. This approach guides existing merging methods by preserving essential layer-wise task-specific knowledge while ensuring the merged model performs the best at benchmarks resembling the calibration dataset. Our experiments demonstrate the effectiveness of LEWIS with performance improvements of code instruction-following and math-solving models created through model merging up to 4 percent and 11.3 percent, respectively, outperforming unguided data-less model merging approaches that use uniform-sparsity.
comment: Accepted at ICLR 2025 Workshop: SLLM (Sparsity in Large Language Models)
☆ Not-Just-Scaling Laws: Towards a Better Understanding of the Downstream Impact of Language Model Design Decisions
Improvements in language model capabilities are often attributed to increasing model size or training data, but in some cases smaller models trained on curated data or with different architectural decisions can outperform larger ones trained on more tokens. What accounts for this? To quantify the impact of these design choices, we meta-analyze 92 open-source pretrained models across a wide array of scales, including state-of-the-art open-weights models as well as less performant models and those with less conventional design decisions. We find that by incorporating features besides model size and number of training tokens, we can achieve a relative 3-28% increase in ability to predict downstream performance compared with using scale alone. Analysis of model design decisions reveal insights into data composition, such as the trade-off between language and code tasks at 15-25\% code, as well as the better performance of some architectural decisions such as choosing rotary over learned embeddings. Broadly, our framework lays a foundation for more systematic investigation of how model development choices shape final capabilities.
☆ Vision-Language Models Struggle to Align Entities across Modalities
Cross-modal entity linking refers to the ability to align entities and their attributes across different modalities. While cross-modal entity linking is a fundamental skill needed for real-world applications such as multimodal code generation, fake news detection, or scene understanding, it has not been thoroughly studied in the literature. In this paper, we introduce a new task and benchmark to address this gap. Our benchmark, MATE, consists of 5.5k evaluation instances featuring visual scenes aligned with their textual representations. To evaluate cross-modal entity linking performance, we design a question-answering task that involves retrieving one attribute of an object in one modality based on a unique attribute of that object in another modality. We evaluate state-of-the-art Vision-Language Models (VLMs) and humans on this task, and find that VLMs struggle significantly compared to humans, particularly as the number of objects in the scene increases. Our analysis also shows that, while chain-of-thought prompting can improve VLM performance, models remain far from achieving human-level proficiency. These findings highlight the need for further research in cross-modal entity linking and show that MATE is a strong benchmark to support that progress.
♻ ☆ MAPS: Motivation-Aware Personalized Search via LLM-Driven Consultation Alignment
Personalized product search aims to retrieve and rank items that match users' preferences and search intent. Despite their effectiveness, existing approaches typically assume that users' query fully captures their real motivation. However, our analysis of a real-world e-commerce platform reveals that users often engage in relevant consultations before searching, indicating they refine intents through consultations based on motivation and need. The implied motivation in consultations is a key enhancing factor for personalized search. This unexplored area comes with new challenges including aligning contextual motivations with concise queries, bridging the category-text gap, and filtering noise within sequence history. To address these, we propose a Motivation-Aware Personalized Search (MAPS) method. It embeds queries and consultations into a unified semantic space via LLMs, utilizes a Mixture of Attention Experts (MoAE) to prioritize critical semantics, and introduces dual alignment: (1) contrastive learning aligns consultations, reviews, and product features; (2) bidirectional attention integrates motivation-aware embeddings with user preferences. Extensive experiments on real and synthetic data show MAPS outperforms existing methods in both retrieval and ranking tasks.
comment: added project repository & dataset URL
♻ ☆ OkraLong: A Flexible Retrieval-Augmented Framework for Long-Text Query Processing
Large Language Models (LLMs) encounter challenges in efficiently processing long-text queries, as seen in applications like enterprise document analysis and financial report comprehension. While conventional solutions employ long-context processing or Retrieval-Augmented Generation (RAG), they suffer from prohibitive input expenses or incomplete information. Recent advancements adopt context compression and dynamic retrieval loops, but still sacrifice critical details or incur iterative costs. To address these limitations, we propose OkraLong, a novel framework that flexibly optimizes the entire processing workflow. Unlike prior static or coarse-grained adaptive strategies, OkraLong adopts fine-grained orchestration through three synergistic components: analyzer, organizer and executor. The analyzer characterizes the task states, which guide the organizer in dynamically scheduling the workflow. The executor carries out the execution and generates the final answer. Experimental results demonstrate that OkraLong not only enhances answer accuracy but also achieves cost-effectiveness across a variety of datasets.
♻ ☆ MCiteBench: A Benchmark for Multimodal Citation Text Generation in MLLMs
Multimodal Large Language Models (MLLMs) have advanced in integrating diverse modalities but frequently suffer from hallucination. A promising solution to mitigate this issue is to generate text with citations, providing a transparent chain for verification. However, existing work primarily focuses on generating citations for text-only content, overlooking the challenges and opportunities of multimodal contexts. To address this gap, we introduce MCiteBench, the first benchmark designed to evaluate and analyze the multimodal citation text generation ability of MLLMs. Our benchmark comprises data derived from academic papers and review-rebuttal interactions, featuring diverse information sources and multimodal content. We comprehensively evaluate models from multiple dimensions, including citation quality, source reliability, and answer accuracy. Through extensive experiments, we observe that MLLMs struggle with multimodal citation text generation. We also conduct deep analyses of models' performance, revealing that the bottleneck lies in attributing the correct sources rather than understanding the multimodal content.
♻ ☆ BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modelling
Time-series Generation (TSG) is a prominent research area with broad applications in simulations, data augmentation, and counterfactual analysis. While existing methods have shown promise in unconditional single-domain TSG, real-world applications demand for cross-domain approaches capable of controlled generation tailored to domain-specific constraints and instance-level requirements. In this paper, we argue that text can provide semantic insights, domain information and instance-specific temporal patterns, to guide and improve TSG. We introduce ``Text-Controlled TSG'', a task focused on generating realistic time series by incorporating textual descriptions. To address data scarcity in this setting, we propose a novel LLM-based Multi-Agent framework that synthesizes diverse, realistic text-to-TS datasets. Furthermore, we introduce BRIDGE, a hybrid text-controlled TSG framework that integrates semantic prototypes with text description for supporting domain-level guidance. This approach achieves state-of-the-art generation fidelity on 11 of 12 datasets, and improves controllability by 12.52% on MSE and 6.34% MAE compared to no text input generation, highlighting its potential for generating tailored time-series data.
comment: Preprint. Work in progress
♻ ☆ SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction ICLR 2025
Large Language Models (LLMs) have demonstrated improved generation performance by incorporating externally retrieved knowledge, a process known as retrieval-augmented generation (RAG). Despite the potential of this approach, existing studies evaluate RAG effectiveness by 1) assessing retrieval and generation components jointly, which obscures retrieval's distinct contribution, or 2) examining retrievers using traditional metrics such as NDCG, which creates a gap in understanding retrieval's true utility in the overall generation process. To address the above limitations, in this work, we introduce an automatic evaluation method that measures retrieval quality through the lens of information gain within the RAG framework. Specifically, we propose Semantic Perplexity (SePer), a metric that captures the LLM's internal belief about the correctness of the retrieved information. We quantify the utility of retrieval by the extent to which it reduces semantic perplexity post-retrieval. Extensive experiments demonstrate that SePer not only aligns closely with human preferences but also offers a more precise and efficient evaluation of retrieval utility across diverse RAG scenarios.
comment: ICLR 2025 Spotlight
♻ ☆ Iterative Value Function Optimization for Guided Decoding
While Reinforcement Learning from Human Feedback (RLHF) has become the predominant method for controlling language model outputs, it suffers from high computational costs and training instability. Guided decoding, especially value-guided methods, offers a cost-effective alternative by controlling outputs without re-training models. However, the accuracy of the value function is crucial for value-guided decoding, as inaccuracies can lead to suboptimal decision-making and degraded performance. Existing methods struggle with accurately estimating the optimal value function, leading to less effective control. We propose Iterative Value Function Optimization, a novel framework that addresses these limitations through two key components: Monte Carlo Value Estimation, which reduces estimation variance by exploring diverse trajectories, and Iterative On-Policy Optimization, which progressively improves value estimation through collecting trajectories from value-guided policies. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of value-guided decoding approaches in aligning language models. These approaches not only achieve alignment but also significantly reduce computational costs by leveraging principled value function optimization for efficient and effective control.
comment: 20 pages, 10 figures
♻ ☆ KiVA: Kid-inspired Visual Analogies for Testing Large Multimodal Models
This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.
comment: 10 pages. Project website: https://ey242.github.io/kiva.github.io/. Benchmark and code: https://github.com/ey242/KiVA
♻ ☆ CPT-Boosted Wav2vec2.0: Towards Noise Robust Speech Recognition for Classroom Environments
Creating Automatic Speech Recognition (ASR) systems that are robust and resilient to classroom conditions is paramount to the development of AI tools to aid teachers and students. In this work, we study the efficacy of continued pretraining (CPT) in adapting Wav2vec2.0 to the classroom domain. We show that CPT is a powerful tool in that regard and reduces the Word Error Rate (WER) of Wav2vec2.0-based models by upwards of 10%. More specifically, CPT improves the model's robustness to different noises, microphones and classroom conditions.
comment: arXiv admin note: substantial text overlap with arXiv:2405.13018
♻ ☆ PARAMANU-GANITA: Can Small Math Language Models Rival with Large Language Models on Mathematical Reasoning?
In this paper, we study whether domain specific pretraining of small generative language models (SLM) from scratch with domain specialized tokenizer and Chain-of-Thought (CoT) instruction fine-tuning results in competitive performance on mathematical reasoning compared to LLMs? Secondly, whether this approach is environmentally sustainable, highly cost efficient? To address these research questions, we present Paramanu-Ganita, a 208 million-parameter novel decoder-only Auto Regressive SLM on mathematics. We performed pretraining from scratch on 31.5 billion tokens for 170 A100 hours using a context size of 4096 on a mixed mathematical corpus consisting of web pages, source code, textbooks, CoT templatised StackOverflow QA pairs, and mathematical lecture notes in LaTeX curated by us. We also trained a math and code specialised BPE tokenizer. We proposed and performed CoT instruction fine-tuning of Paramanu-Ganita on the MetaMathQA dataset. Our model Paramanu-Ganita, despite being 34 times smaller than the 7B LLMs, outperforms generalist LLMs by approximately 30% points, and even math-specialised LLMs by 3-23% points in GSM8K test accuracy metric. On MATH benchmark, Paramanu-Ganita outperformed the various models by 6-8% points. On benchmarks like LogiQA, MMLU (high school, college level), and competitive exams level, AGIEVAL (AQuA-RAT, SAT-Math), Paramanu-Ganita outperformed others by 1-4%. Our model is available at https://huggingface.co/gyanai/paramanu-ganita-208M-hf .
♻ ☆ DelTA: An Online Document-Level Translation Agent Based on Multi-Level Memory ICLR 2025
Large language models (LLMs) have achieved reasonable quality improvements in machine translation (MT). However, most current research on MT-LLMs still faces significant challenges in maintaining translation consistency and accuracy when processing entire documents. In this paper, we introduce DelTA, a Document-levEL Translation Agent designed to overcome these limitations. DelTA features a multi-level memory structure that stores information across various granularities and spans, including Proper Noun Records, Bilingual Summary, Long-Term Memory, and Short-Term Memory, which are continuously retrieved and updated by auxiliary LLM-based components. Experimental results indicate that DelTA significantly outperforms strong baselines in terms of translation consistency and quality across four open/closed-source LLMs and two representative document translation datasets, achieving an increase in consistency scores by up to 4.58 percentage points and in COMET scores by up to 3.16 points on average. DelTA employs a sentence-by-sentence translation strategy, ensuring no sentence omissions and offering a memory-efficient solution compared to the mainstream method. Furthermore, DelTA improves pronoun and context-dependent translation accuracy, and the summary component of the agent also shows promise as a tool for query-based summarization tasks. The code and data of our approach are released at https://github.com/YutongWang1216/DocMTAgent.
comment: Accepted as a conference paper at ICLR 2025
♻ ☆ CTC-DRO: Robust Optimization for Reducing Language Disparities in Speech Recognition
Modern deep learning models often achieve high overall performance, but consistently fail on specific subgroups. Group distributionally robust optimization (group DRO) addresses this problem by minimizing the worst-group loss, but it fails when group losses misrepresent performance differences between groups. This is common in domains like speech, where the widely used connectionist temporal classification (CTC) loss scales with input length and varies with linguistic and acoustic properties, leading to spurious differences between group losses. We present CTC-DRO, which addresses the shortcomings of the group DRO objective by smoothing the group weight update to prevent overemphasis on consistently high-loss groups, while using input length-matched batching to mitigate CTC's scaling issues. We evaluate CTC-DRO on the task of multilingual automatic speech recognition (ASR) across five language sets from the ML-SUPERB 2.0 benchmark. CTC-DRO consistently outperforms group DRO and CTC-based baseline models, reducing the worst-language error by up to 47.1% and the average error by up to 32.9%. CTC-DRO can be applied to ASR with minimal computational costs, and offers the potential for reducing group disparities in other domains with similar challenges.
♻ ☆ What to align in multimodal contrastive learning? ICLR 2025
Humans perceive the world through multisensory integration, blending the information of different modalities to adapt their behavior. Contrastive learning offers an appealing solution for multimodal self-supervised learning. Indeed, by considering each modality as a different view of the same entity, it learns to align features of different modalities in a shared representation space. However, this approach is intrinsically limited as it only learns shared or redundant information between modalities, while multimodal interactions can arise in other ways. In this work, we introduce CoMM, a Contrastive MultiModal learning strategy that enables the communication between modalities in a single multimodal space. Instead of imposing cross- or intra- modality constraints, we propose to align multimodal representations by maximizing the mutual information between augmented versions of these multimodal features. Our theoretical analysis shows that shared, synergistic and unique terms of information naturally emerge from this formulation, allowing us to estimate multimodal interactions beyond redundancy. We test CoMM both in a controlled and in a series of real-world settings: in the former, we demonstrate that CoMM effectively captures redundant, unique and synergistic information between modalities. In the latter, CoMM learns complex multimodal interactions and achieves state-of-the-art results on the seven multimodal benchmarks. Code is available at https://github.com/Duplums/CoMM
comment: ICLR 2025, 25 pages
♻ ☆ CycleResearcher: Improving Automated Research via Automated Review ICLR 2025
The automation of scientific discovery has been a long-standing goal within the research community, driven by the potential to accelerate knowledge creation. While significant progress has been made using commercial large language models (LLMs) as research assistants or idea generators, the possibility of automating the entire research process with open-source LLMs remains largely unexplored. This paper explores the feasibility of using open-source post-trained LLMs as autonomous agents capable of performing the full cycle of automated research and review, from literature review and manuscript preparation to peer review and paper refinement. Our iterative preference training framework consists of CycleResearcher, which conducts research tasks, and CycleReviewer, which simulates the peer review process, providing iterative feedback via reinforcement learning. To train these models, we develop two new datasets, Review-5k and Research-14k, reflecting real-world machine learning research and peer review dynamics. Our results demonstrate that CycleReviewer achieves promising performance with a 26.89\% reduction in mean absolute error (MAE) compared to individual human reviewers in predicting paper scores, indicating the potential of LLMs to effectively assist expert-level research evaluation. In research, the papers generated by the CycleResearcher model achieved a score of 5.36 in simulated peer reviews, showing some competitiveness in terms of simulated review scores compared to the preprint level of 5.24 from human experts, while still having room for improvement compared to the accepted paper level of 5.69. This work represents a significant step toward fully automated scientific inquiry, providing ethical safeguards and exploring AI-driven research capabilities. The code, dataset and model weight are released at https://wengsyx.github.io/Researcher/
comment: Accept in ICLR 2025
♻ ☆ The Noisy Path from Source to Citation: Measuring How Scholars Engage with Past Research
Academic citations are widely used for evaluating research and tracing knowledge flows. Such uses typically rely on raw citation counts and neglect variability in citation types. In particular, citations can vary in their fidelity as original knowledge from cited studies may be paraphrased, summarized, or reinterpreted, possibly wrongly, leading to variation in how much information changes from cited to citing paper. In this study, we introduce a computational pipeline to quantify citation fidelity at scale. Using full texts of papers, the pipeline identifies citations in citing papers and the corresponding claims in cited papers, and applies supervised models to measure fidelity at the sentence level. Analyzing a large-scale multi-disciplinary dataset of approximately 13 million citation sentence pairs, we find that citation fidelity is higher when authors cite papers that are 1) more recent and intellectually close, 2) more accessible, and 3) the first author has a lower H-index and the author team is medium-sized. Using a quasi-experiment, we establish the "telephone effect" - when citing papers have low fidelity to the original claim, future papers that cite the citing paper and the original have lower fidelity to the original. Our work reveals systematic differences in citation fidelity, underscoring the limitations of analyses that rely on citation quantity alone and the potential for distortion of evidence.
♻ ☆ Towards Effective Time-Aware Language Representation: Exploring Enhanced Temporal Understanding in Language Models
In the evolving field of Natural Language Processing (NLP), understanding the temporal context of text is increasingly critical for applications requiring advanced temporal reasoning. Traditional pre-trained language models like BERT, which rely on synchronic document collections such as BookCorpus and Wikipedia, often fall short in effectively capturing and leveraging temporal information. To address this limitation, we introduce BiTimeBERT 2.0, a novel time-aware language model pre-trained on a temporal news article collection. BiTimeBERT 2.0 incorporates temporal information through three innovative pre-training objectives: Extended Time-Aware Masked Language Modeling (ETAMLM), Document Dating (DD), and Time-Sensitive Entity Replacement (TSER). Each objective is specifically designed to target a distinct dimension of temporal information: ETAMLM enhances the model's understanding of temporal contexts and relations, DD integrates document timestamps as explicit chronological markers, and TSER focuses on the temporal dynamics of "Person" entities. Moreover, our refined corpus preprocessing strategy reduces training time by nearly 53\%, making BiTimeBERT 2.0 significantly more efficient while maintaining high performance. Experimental results show that BiTimeBERT 2.0 achieves substantial improvements across a broad range of time-related tasks and excels on datasets spanning extensive temporal ranges. These findings underscore BiTimeBERT 2.0's potential as a powerful tool for advancing temporal reasoning in NLP.
comment: This paper has been accepted for publication in ACM Transactions on the Web. Final publication details (volume, issue, page range) will be updated once they are finalized
♻ ☆ FairSense-AI: Responsible AI Meets Sustainability
In this paper, we introduce FairSense-AI: a multimodal framework designed to detect and mitigate bias in both text and images. By leveraging Large Language Models (LLMs) and Vision-Language Models (VLMs), FairSense-AI uncovers subtle forms of prejudice or stereotyping that can appear in content, providing users with bias scores, explanatory highlights, and automated recommendations for fairness enhancements. In addition, FairSense-AI integrates an AI risk assessment component that aligns with frameworks like the MIT AI Risk Repository and NIST AI Risk Management Framework, enabling structured identification of ethical and safety concerns. The platform is optimized for energy efficiency via techniques such as model pruning and mixed-precision computation, thereby reducing its environmental footprint. Through a series of case studies and applications, we demonstrate how FairSense-AI promotes responsible AI use by addressing both the social dimension of fairness and the pressing need for sustainability in large-scale AI deployments. https://vectorinstitute.github.io/FairSense-AI, https://pypi.org/project/fair-sense-ai/ (Sustainability , Responsible AI , Large Language Models , Vision Language Models , Ethical AI , Green AI)
♻ ☆ Provable Benefits of Task-Specific Prompts for In-context Learning AISTATS
The in-context learning capabilities of modern language models have motivated a deeper mathematical understanding of sequence models. A line of recent work has shown that linear attention models can emulate projected gradient descent iterations to implicitly learn the task vector from the data provided in the context window. In this work, we consider a novel setting where the global task distribution can be partitioned into a union of conditional task distributions. We then examine the use of task-specific prompts and prediction heads for learning the prior information associated with the conditional task distribution using a one-layer attention model. Our results on loss landscape show that task-specific prompts facilitate a covariance-mean decoupling where prompt-tuning explains the conditional mean of the distribution whereas the variance is learned/explained through in-context learning. Incorporating task-specific head further aids this process by entirely decoupling estimation of mean and variance components. This covariance-mean perspective similarly explains how jointly training prompt and attention weights can provably help over fine-tuning after pretraining.
comment: Proceedings of the 28th International Conference on Artificial Intelligence and Statistics (AISTATS) 2025
♻ ☆ MIRROR: A Novel Approach for the Automated Evaluation of Open-Ended Question Generation NeurIPS'24
Automatic question generation is a critical task that involves evaluating question quality by considering factors such as engagement, pedagogical value, and the ability to stimulate critical thinking. These aspects require human-like understanding and judgment, which automated systems currently lack. However, human evaluations are costly and impractical for large-scale samples of generated questions. Therefore, we propose a novel system, MIRROR (Multi-LLM Iterative Review and Response for Optimized Rating), which leverages large language models (LLMs) to automate the evaluation process for questions generated by automated question generation systems. We experimented with several state-of-the-art LLMs, such as GPT-4, Gemini, and Llama2-70b. We observed that the scores of human evaluation metrics, namely relevance, appropriateness, novelty, complexity, and grammaticality, improved when using the feedback-based approach called MIRROR, tending to be closer to the human baseline scores. Furthermore, we observed that Pearson's correlation coefficient between GPT-4 and human experts improved when using our proposed feedback-based approach, MIRROR, compared to direct prompting for evaluation. Error analysis shows that our proposed approach, MIRROR, significantly helps to improve relevance and appropriateness.
comment: NeurIPS'24 Workshop on Large Foundation Models for Educational Assessment (FM-EduAssess)
♻ ☆ Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification
The ability to process long contexts is crucial for many natural language processing tasks, yet it remains a significant challenge. While substantial progress has been made in enhancing the efficiency of attention mechanisms, there is still a gap in understanding how attention heads function in long-context settings. In this paper, we observe that while certain heads consistently attend to local information only, others swing between attending to local and long-context information depending on the query. This raises the question: can we identify which heads require long-context information to predict the next token accurately? We demonstrate that it's possible to predict which heads are crucial for long-context processing using only local keys. The core idea here is to exploit a simple model for the long-context scores via second moment approximations. These findings unveil simple properties of attention in the context of long sequences, and open the door to potentially significant gains in efficiency.
♻ ☆ More than Memes: A Multimodal Topic Modeling Approach to Conspiracy Theories on Telegram
To address the increasing prevalence of (audio-)visual data on social media, and to capture the evolving and dynamic nature of this communication, researchers have begun to explore the potential of unsupervised approaches for analyzing multimodal online content. However, existing research often neglects visual content beyond memes, and in addition lacks methods to compare topic models across modalities. Our study addresses these gaps by applying multimodal topic modeling for analyzing conspiracy theories in German-language Telegram channels. We use BERTopic with CLIP for the analysis of textual and visual data in a corpus of ~40, 000 Telegram messages posted in October 2023 in 571 German-language Telegram channels known for disseminating conspiracy theories. Through this dataset, we provide insights into unimodal and multimodal topic models by analyzing symmetry and intersections of topics across modalities. We demonstrate the variety of textual and visual content shared in the channels discovered through the topic modeling, and propose a conceptual framework for the analysis of textual and visual discursive strategies in the communication of conspiracy theories. We apply the framework in a case study of the topic group Israel Gaza.
comment: 12 pages, 10 figures
♻ ☆ Prompt-enhanced Network for Hateful Meme Classification
The dynamic expansion of social media has led to an inundation of hateful memes on media platforms, accentuating the growing need for efficient identification and removal. Acknowledging the constraints of conventional multimodal hateful meme classification, which heavily depends on external knowledge and poses the risk of including irrelevant or redundant content, we developed Pen -- a prompt-enhanced network framework based on the prompt learning approach. Specifically, after constructing the sequence through the prompt method and encoding it with a language model, we performed region information global extraction on the encoded sequence for multi-view perception. By capturing global information about inference instances and demonstrations, Pen facilitates category selection by fully leveraging sequence information. This approach significantly improves model classification accuracy. Additionally, to bolster the model's reasoning capabilities in the feature space, we introduced prompt-aware contrastive learning into the framework to improve the quality of sample feature distributions. Through extensive ablation experiments on two public datasets, we evaluate the effectiveness of the Pen framework, concurrently comparing it with state-of-the-art model baselines. Our research findings highlight that Pen surpasses manual prompt methods, showcasing superior generalization and classification accuracy in hateful meme classification tasks. Our code is available at https://github.com/juszzi/Pen.
comment: Published in Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence Main Track. Pages 6397-6405
♻ ☆ From Informal to Formal -- Incorporating and Evaluating LLMs on Natural Language Requirements to Verifiable Formal Proofs
The research in AI-based formal mathematical reasoning has shown an unstoppable growth trend. These studies have excelled in mathematical competitions like IMO and have made significant progress. This paper focuses on formal verification, an immediate application scenario of formal reasoning, and breaks it down into sub-tasks. We constructed 18k high-quality instruction-response pairs across five formal specification languages (Coq, Lean4, Dafny, ACSL, and TLA+) by distilling gpt-4o and evaluated against ten open-sourced LLMs, including recent popular DeepSeek-R1. We also fine-tuned several 7~8B small models to achieve comparable performance with Deepseek-R1-671B. Interestingly, we observed that fine-tuning with formal data also enhances mathematics, reasoning, and coding capabilities. Fine-tuned models are released at https: //huggingface.co/fm-universe.
comment: 19 pages
♻ ☆ Protecting multimodal large language models against misleading visualizations
We assess the vulnerability of multimodal large language models to misleading visualizations - charts that distort the underlying data using techniques such as truncated or inverted axes, leading readers to draw inaccurate conclusions that may support misinformation or conspiracy theories. Our analysis shows that these distortions severely harm multimodal large language models, reducing their question-answering accuracy to the level of the random baseline. To mitigate this vulnerability, we introduce six inference-time methods to improve performance of MLLMs on misleading visualizations while preserving their accuracy on non-misleading ones. The most effective approach involves (1) extracting the underlying data table and (2) using a text-only large language model to answer questions based on the table. This method improves performance on misleading visualizations by 15.4 to 19.6 percentage points.
comment: Preprint. Code and data available at https://github.com/UKPLab/arxiv2025-misleading-visualizations
♻ ☆ Rewarding Doubt: A Reinforcement Learning Approach to Confidence Calibration of Large Language Models
A safe and trustworthy use of Large Language Models (LLMs) requires an accurate expression of confidence in their answers. We introduce a novel Reinforcement Learning (RL) approach for LLM calibration that fine-tunes LLMs to elicit calibrated confidence estimations in their answers to factual questions. We model the problem as a betting game where the model predicts a confidence score together with every answer, and design a reward function that penalizes both over and under-confidence. We prove that under our reward design an optimal policy would result in a perfectly calibrated confidence estimation. Our experiments demonstrate significantly improved confidence calibration and generalization to new tasks without re-training, indicating that our approach teaches a general confidence awareness. This approach enables the training of inherently calibrated LLMs.
♻ ☆ LLMs can be Dangerous Reasoners: Analyzing-based Jailbreak Attack on Large Language Models
The rapid development of Large Language Models (LLMs) has brought significant advancements across various tasks. However, despite these achievements, LLMs still exhibit inherent safety vulnerabilities, especially when confronted with jailbreak attacks. Existing jailbreak methods suffer from two main limitations: reliance on complicated prompt engineering and iterative optimization, which lead to low attack success rate (ASR) and attack efficiency (AE). In this work, we propose an efficient jailbreak attack method, Analyzing-based Jailbreak (ABJ), which leverages the advanced reasoning capability of LLMs to autonomously generate harmful content, revealing their underlying safety vulnerabilities during complex reasoning process. We conduct comprehensive experiments on ABJ across various open-source and closed-source LLMs. In particular, ABJ achieves high ASR (82.1% on GPT-4o-2024-11-20) with exceptional AE among all target LLMs, showcasing its remarkable attack effectiveness, transferability, and efficiency. Our findings underscore the urgent need to prioritize and improve the safety of LLMs to mitigate the risks of misuse.
♻ ☆ Building Safe GenAI Applications: An End-to-End Overview of Red Teaming for Large Language Models
The rapid growth of Large Language Models (LLMs) presents significant privacy, security, and ethical concerns. While much research has proposed methods for defending LLM systems against misuse by malicious actors, researchers have recently complemented these efforts with an offensive approach that involves red teaming, i.e., proactively attacking LLMs with the purpose of identifying their vulnerabilities. This paper provides a concise and practical overview of the LLM red teaming literature, structured so as to describe a multi-component system end-to-end. To motivate red teaming we survey the initial safety needs of some high-profile LLMs, and then dive into the different components of a red teaming system as well as software packages for implementing them. We cover various attack methods, strategies for attack-success evaluation, metrics for assessing experiment outcomes, as well as a host of other considerations. Our survey will be useful for any reader who wants to rapidly obtain a grasp of the major red teaming concepts for their own use in practical applications.
♻ ☆ RIDE: Enhancing Large Language Model Alignment through Restyled In-Context Learning Demonstration Exemplars
Alignment tuning is crucial for ensuring large language models (LLMs) behave ethically and helpfully. Current alignment approaches require high-quality annotations and significant training resources. This paper proposes a low-cost, tuning-free method using in-context learning (ICL) to enhance LLM alignment. Through an analysis of high-quality ICL demos, we identified style as a key factor influencing LLM alignment capabilities and explicitly restyled ICL exemplars based on this stylistic framework. Additionally, we combined the restyled demos to achieve a balance between the two conflicting aspects of LLM alignment--factuality and safety. We packaged the restyled examples as prompts to trigger few-shot learning, improving LLM alignment. Compared to the best baseline approach, with an average score of 5.00 as the maximum, our method achieves a maximum 0.10 increase on the Alpaca task (from 4.50 to 4.60), a 0.22 enhancement on the Just-eval benchmark (from 4.34 to 4.56), and a maximum improvement of 0.32 (from 3.53 to 3.85) on the MT-Bench dataset. We release the code and data at https://github.com/AnonymousCode-ComputerScience/RIDE.
comment: 38 pages, 2 figures, 20 tables; The paper is under review in ARR
♻ ☆ What is a Digital Twin Anyway? Deriving the Definition for the Built Environment from over 15,000 Scientific Publications
The concept of digital twins has attracted significant attention across various domains, particularly within the built environment. However, there is a sheer volume of definitions and the terminological consensus remains out of reach. The lack of a universally accepted definition leads to ambiguities in their conceptualization and implementation, and may cause miscommunication for both researchers and practitioners. We employed Natural Language Processing (NLP) techniques to systematically extract and analyze definitions of digital twins from a corpus of more than 15,000 full-text articles spanning diverse disciplines. The study compares these findings with insights from an expert survey that included 52 experts. The study identifies concurrence on the components that comprise a ``Digital Twin'' from a practical perspective across various domains, contrasting them with those that do not, to identify deviations. We investigate the evolution of digital twin definitions over time and across different scales, including manufacturing, building, and urban/geospatial perspectives. We extracted the main components of Digital Twins using Text Frequency Analysis and N-gram analysis. Subsequently, we identified components that appeared in the literature and conducted a Chi-square test to assess the significance of each component in different domains. Our analysis identified key components of digital twins and revealed significant variations in definitions based on application domains, such as manufacturing, building, and urban contexts. The analysis of DT components reveal two major groups of DT types: High-Performance Real-Time (HPRT) DTs, and Long-Term Decision Support (LTDS) DTs. Contrary to common assumptions, we found that components such as simulation, AI/ML, real-time capabilities, and bi-directional data flow are not yet fully mature in the digital twins of the built environment.
♻ ☆ From Sparse Dependence to Sparse Attention: Unveiling How Chain-of-Thought Enhances Transformer Sample Efficiency
Chain-of-thought (CoT) significantly enhances the reasoning performance of large language models (LLM). While current theoretical studies often attribute this improvement to increased expressiveness and computational capacity, we argue that expressiveness is not the primary limitation in the LLM regime, as current large models will fail on simple tasks. Using a parity-learning setup, we demonstrate that CoT can substantially improve sample efficiency even when the representation power is sufficient. Specifically, with CoT, a transformer can learn the function within polynomial samples, whereas without CoT, the required sample size is exponential. Additionally, we show that CoT simplifies the learning process by introducing sparse sequential dependencies among input tokens, and leads to a sparse and interpretable attention. We validate our theoretical analysis with both synthetic and real-world experiments, confirming that sparsity in attention layers is a key factor of the improvement induced by CoT.
comment: 43 pages,11 figures
♻ ☆ Promote, Suppress, Iterate: How Language Models Answer One-to-Many Factual Queries
To answer one-to-many factual queries (e.g., listing cities of a country), a language model (LM) must simultaneously recall knowledge and avoid repeating previous answers. How are these two subtasks implemented and integrated internally? Across multiple datasets and models, we identify a promote-then-suppress mechanism: the model first recalls all answers, and then suppresses previously generated ones. Specifically, LMs use both the subject and previous answer tokens to perform knowledge recall, with attention propagating subject information and MLPs promoting the answers. Then, attention attends to and suppresses previous answer tokens, while MLPs amplify the suppression signal. Our mechanism is corroborated by extensive experimental evidence: in addition to using early decoding and causal tracing, we analyze how components use different tokens by introducing both Token Lens, which decodes aggregated attention updates from specified tokens, and a knockout method that analyzes changes in MLP outputs after removing attention to specified tokens. Overall, we provide new insights into how LMs' internal components interact with different input tokens to support complex factual recall. Code is available at https://github.com/Lorenayannnnn/how-lms-answer-one-to-many-factual-queries.
♻ ☆ Graph-Aware Isomorphic Attention for Adaptive Dynamics in Transformers
We present an approach to modifying Transformer architectures by integrating graph-aware relational reasoning into the attention mechanism, merging concepts from graph neural networks and language modeling. Building on the inherent connection between attention and graph theory, we reformulate the Transformer's attention mechanism as a graph operation and propose Graph-Aware Isomorphic Attention. This method leverages advanced graph modeling strategies, including Graph Isomorphism Networks (GIN) and Principal Neighborhood Aggregation (PNA), to enrich the representation of relational structures. Our approach captures complex dependencies and generalizes across tasks, as evidenced by a reduced generalization gap and improved learning performance. Additionally, we expand the concept of graph-aware attention to introduce Sparse GIN-Attention, a fine-tuning approach that employs sparse GINs. By interpreting attention matrices as sparse adjacency graphs, this technique enhances the adaptability of pre-trained foundational models with minimal computational overhead, endowing them with graph-aware capabilities. Sparse GIN-Attention fine-tuning achieves improved training dynamics and better generalization compared to alternative methods like low-rank adaption (LoRA). We discuss latent graph-like structures within traditional attention mechanisms, offering a new lens through which Transformers can be understood. By evolving Transformers as hierarchical GIN models for relational reasoning. This perspective suggests profound implications for foundational model development, enabling the design of architectures that dynamically adapt to both local and global dependencies. Applications in bioinformatics, materials science, language modeling, and beyond could benefit from this synthesis of relational and sequential data modeling, setting the stage for interpretable and generalizable modeling strategies.
♻ ☆ Enhancing Non-English Capabilities of English-Centric Large Language Models through Deep Supervision Fine-Tuning AAAI 2025
Large language models (LLMs) have demonstrated significant progress in multilingual language understanding and generation. However, due to the imbalance in training data, their capabilities in non-English languages are limited. Recent studies revealed the English-pivot multilingual mechanism of LLMs, where LLMs implicitly convert non-English queries into English ones at the bottom layers and adopt English for thinking at the middle layers. However, due to the absence of explicit supervision for cross-lingual alignment in the intermediate layers of LLMs, the internal representations during these stages may become inaccurate. In this work, we introduce a deep supervision fine-tuning method (DFT) that incorporates additional supervision in the internal layers of the model to guide its workflow. Specifically, we introduce two training objectives on different layers of LLMs: one at the bottom layers to constrain the conversion of the target language into English, and another at the middle layers to constrain reasoning in English. To effectively achieve the guiding purpose, we designed two types of supervision signals: logits and feature, which represent a stricter constraint and a relatively more relaxed guidance. Our method guides the model to not only consider the final generated result when processing non-English inputs but also ensure the accuracy of internal representations. We conducted extensive experiments on typical English-centric large models, LLaMA-2 and Gemma-2, and the results on multiple multilingual datasets show that our method significantly outperforms traditional fine-tuning methods.
comment: Accepted at AAAI 2025
♻ ☆ ChaI-TeA: A Benchmark for Evaluating Autocompletion of Interactions with LLM-based Chatbots
The rise of LLMs has deflected a growing portion of human-computer interactions towards LLM-based chatbots. The remarkable abilities of these models allow users to interact using long, diverse natural language text covering a wide range of topics and styles. Phrasing these messages is a time and effort consuming task, calling for an autocomplete solution to assist users. We introduce the task of chatbot interaction autocomplete. We present ChaI-TeA: CHat InTEraction Autocomplete; An autcomplete evaluation framework for LLM-based chatbot interactions. The framework includes a formal definition of the task, coupled with suitable datasets and metrics. We use the framework to evaluate After formally defining the task along with suitable datasets and metrics, we test 9 models on the defined auto completion task, finding that while current off-the-shelf models perform fairly, there is still much room for improvement, mainly in ranking of the generated suggestions. We provide insights for practitioners working on this task and open new research directions for researchers in the field. We release our framework to serve as a foundation for future research.
♻ ☆ Explaining Vision-Language Similarities in Dual Encoders with Feature-Pair Attributions
Dual encoder architectures like CLIP models map two types of inputs into a shared embedding space and predict similarities between them. Despite their success, it is, however, not understood how these models compare their two inputs. Common first-order feature-attribution methods can only provide limited insights into dual-encoders since their predictions depend on feature-interactions rather than on individual features. In this paper, we first derive a second-order method enabling the attribution of predictions by any differentiable dual encoder onto feature-interactions between its inputs. Second, we apply our method to CLIP models and show that they learn fine-grained correspondences between parts of captions and regions in images. They match objects across input modes also account for mismatches. This visual-linguistic grounding ability, however, varies heavily between object classes and exhibits pronounced out-of-domain effects. We can identify individual errors as well as systematic failure categories including object coverage, unusual scenes and correlated contexts.
♻ ☆ Number Cookbook: Number Understanding of Language Models and How to Improve It ICLR 2025
Large language models (LLMs) can solve an increasing number of complex reasoning tasks while making surprising mistakes in basic numerical understanding and processing (such as 9.11 > 9.9). The latter ability is essential for tackling complex arithmetic and mathematical problems and serves as a foundation for most reasoning tasks, but previous work paid little attention to it or only discussed several restricted tasks (like integer addition). In this paper, we comprehensively investigate the numerical understanding and processing ability (NUPA) of LLMs. Firstly, we introduce a benchmark covering four common numerical representations and 17 distinct numerical tasks in four major categories, resulting in 41 meaningful combinations in total. These tasks are derived from primary and secondary education curricula, encompassing nearly all everyday numerical understanding and processing scenarios, and the rules of these tasks are very simple and clear. Through the benchmark, we find that current LLMs fail frequently in many of the tasks. To study the problem, we train small models with existing and potential techniques for enhancing NUPA (such as tokenizers, PEs, and number formats), comprehensively evaluating their effectiveness using our testbed. We also finetune practical-scale LLMs on our proposed NUPA tasks and find that 1) naive finetuning can improve NUPA a lot on many but not all tasks, and 2) surprisingly, techniques designed to enhance NUPA prove ineffective for finetuning pretrained models. We further explore the impact of chain-of-thought techniques on NUPA. Our work provides a more detailed and comprehensive understanding of NUPA in LLMs. Our benchmark and code are released at https://github.com/GraphPKU/number_cookbook.
comment: ICLR 2025 poster
♻ ☆ DarwinLM: Evolutionary Structured Pruning of Large Language Models
Large Language Models (LLMs) have achieved significant success across various NLP tasks. However, their massive computational costs limit their widespread use, particularly in real-time applications. Structured pruning offers an effective solution by compressing models and directly providing end-to-end speed improvements, regardless of the hardware environment. Meanwhile, different components of the model exhibit varying sensitivities towards pruning, calling for non-uniform model compression. However, a pruning method should not only identify a capable substructure, but also account for post-compression training. To this end, we propose DarwinLM, a method for training-aware structured pruning. DarwinLM builds upon an evolutionary search process, generating multiple offspring models in each generation through mutation, and selecting the fittest for survival. To assess the effect of post-training, we incorporate a lightweight, multistep training process within the offspring population, progressively increasing the number of tokens and eliminating poorly performing models in each selection stage. We validate our method through extensive experiments on Llama-2-7B, Llama-3.1-8B and Qwen-2.5-14B-Instruct, achieving state-of-the-art performance for structured pruning. For instance, DarwinLM surpasses ShearedLlama while requiring 5x less training data during post-compression training. Code is at: https://github.com/IST-DASLab/DarwinLM
comment: Code: https://github.com/IST-DASLab/DarwinLM
♻ ☆ DuplexMamba: Enhancing Real-time Speech Conversations with Duplex and Streaming Capabilities
Real-time speech conversation is essential for natural and efficient human-machine interactions, requiring duplex and streaming capabilities. Traditional Transformer-based conversational chatbots operate in a turn-based manner and exhibit quadratic computational complexity that grows as the input size increases. In this paper, we propose DuplexMamba, a Mamba-based end-to-end multimodal duplex model for speech-to-text conversation. DuplexMamba enables simultaneous input processing and output generation, dynamically adjusting to support real-time streaming. Specifically, we develop a Mamba-based speech encoder and adapt it with a Mamba-based language model. Furthermore, we introduce a novel duplex decoding strategy that enables DuplexMamba to process input and generate output simultaneously. Experimental results demonstrate that DuplexMamba successfully implements duplex and streaming capabilities while achieving performance comparable to several recently developed Transformer-based models in automatic speech recognition (ASR) tasks and voice assistant benchmark evaluations. Our code and model are released
comment: 12 pages, 6 figures
♻ ☆ On-Policy Self-Alignment with Fine-grained Knowledge Feedback for Hallucination Mitigation
Hallucination occurs when large language models exhibit behavior that deviates from the boundaries of their knowledge during response generation. To address this critical issue, previous learning-based methods attempt to finetune models but are limited by off-policy sampling and coarse-grained feedback. In this paper, we present \textit{\b{R}einforcement \b{L}earning \b{f}or \b{H}allucination} (RLFH), an on-policy self-alignment approach that enables LLMs to actively explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals. RLFH introduces a self-assessment framework where the policy serves as its own judge. Through this framework, responses are automatically decomposed into atomic facts and their truthfulness and informativeness are assessed against external knowledge sources. The resulting fine-grained feedback at the statement level are then converted into token-level dense reward signals. This enables online reinforcement learning to achieve precise and timely optimization without human intervention. Comprehensive evaluations on HotpotQA, SQuADv2, and Biography benchmarks validate RLFH's effectiveness in hallucination mitigation.
♻ ☆ Affordably Fine-tuned LLMs Provide Better Answers to Course-specific MCQs
In education, the capability of generating human-like text of Large Language Models (LLMs) inspired work on how they can increase the efficiency of learning and teaching. We study the affordability of these models for educators and students by investigating how LLMs answer multiple-choice questions (MCQs) with respect to hardware constraints and refinement techniques. We explore this space by using generic pre-trained LLMs (the 7B, 13B, and 70B variants of LLaMA-2) to answer 162 undergraduate-level MCQs from a course on Programming Languages (PL) -- the MCQ dataset is a contribution of this work, which we make publicly available. Specifically, we dissect how different factors, such as using readily-available material -- (parts of) the course's textbook -- for fine-tuning and quantisation (to decrease resource usage) can change the accuracy of the responses. The main takeaway is that smaller textbook-based fine-tuned models outperform generic larger ones (whose pre-training requires conspicuous resources), making the usage of LLMs for answering MCQs resource- and material-wise affordable.
comment: The 40th ACM/SIGAPP Symposium On Applied Computing
♻ ☆ LaRA: Benchmarking Retrieval-Augmented Generation and Long-Context LLMs -- No Silver Bullet for LC or RAG Routing
Effectively incorporating external knowledge into Large Language Models (LLMs) is crucial for enhancing their capabilities and addressing real-world needs. Retrieval-Augmented Generation (RAG) offers an effective method for achieving this by retrieving the most relevant fragments into LLMs. However, the advancements in context window size for LLMs offer an alternative approach, raising the question of whether RAG remains necessary for effectively handling external knowledge. Several existing studies provide inconclusive comparisons between RAG and long-context (LC) LLMs, largely due to limitations in the benchmark designs. In this paper, we present LaRA, a novel benchmark specifically designed to rigorously compare RAG and LC LLMs. LaRA encompasses 2326 test cases across four practical QA task categories and three types of naturally occurring long texts. Through systematic evaluation of seven open-source and four proprietary LLMs, we find that the optimal choice between RAG and LC depends on a complex interplay of factors, including the model's parameter size, long-text capabilities, context length, task type, and the characteristics of the retrieved chunks. Our findings provide actionable guidelines for practitioners to effectively leverage both RAG and LC approaches in developing and deploying LLM applications. Our code and dataset is provided at: \href{https://github.com/Alibaba-NLP/LaRA}{\textbf{https://github.com/Alibaba-NLP/LaRA}}.
comment: 22 pages
♻ ☆ Gated Delta Networks: Improving Mamba2 with Delta Rule ICLR 2025
Linear Transformers have gained attention as efficient alternatives to standard Transformers, but their performance in retrieval and long-context tasks has been limited. To address these limitations, recent work has explored two distinct mechanisms: gating for adaptive memory control and the delta update rule for precise memory modifications. We observe that these mechanisms are complementary: gating enables rapid memory erasure while the delta rule facilitates targeted updates. Building on this insight, we introduce the gated delta rule and develop a parallel training algorithm optimized for modern hardware. Our proposed architecture, Gated DeltaNet, consistently surpasses existing models like Mamba2 and DeltaNet across multiple benchmarks, including language modeling, common-sense reasoning, in-context retrieval, length extrapolation, and long-context understanding. We further enhance performance by developing hybrid architectures that combine Gated DeltaNet layers with sliding window attention or Mamba2 layers, achieving both improved training efficiency and superior task performance.
comment: ICLR 2025 camera ready
♻ ☆ Baichuan-M1: Pushing the Medical Capability of Large Language Models
The current generation of large language models (LLMs) is typically designed for broad, general-purpose applications, while domain-specific LLMs, especially in vertical fields like medicine, remain relatively scarce. In particular, the development of highly efficient and practical LLMs for the medical domain is challenging due to the complexity of medical knowledge and the limited availability of high-quality data. To bridge this gap, we introduce Baichuan-M1, a series of large language models specifically optimized for medical applications. Unlike traditional approaches that simply continue pretraining on existing models or apply post-training to a general base model, Baichuan-M1 is trained from scratch with a dedicated focus on enhancing medical capabilities. Our model is trained on 20 trillion tokens and incorporates a range of effective training methods that strike a balance between general capabilities and medical expertise. As a result, Baichuan-M1 not only performs strongly across general domains such as mathematics and coding but also excels in specialized medical fields. We have open-sourced Baichuan-M1-14B, a mini version of our model, which can be accessed through the following links.
comment: 33 pages, technical report
♻ ☆ StableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of Large Language Models
Large Language Models (LLMs) have witnessed remarkable advancements in recent years, prompting the exploration of tool learning, which integrates LLMs with external tools to address diverse real-world challenges. Assessing the capability of LLMs to utilise tools necessitates large-scale and stable benchmarks. However, previous works relied on either hand-crafted online tools with limited scale, or large-scale real online APIs suffering from instability of API status. To address this problem, we introduce StableToolBench, a benchmark evolving from ToolBench, proposing a virtual API server and stable evaluation system. The virtual API server contains a caching system and API simulators which are complementary to alleviate the change in API status. Meanwhile, the stable evaluation system designs solvable pass and win rates using GPT-4 as the automatic evaluator to eliminate the randomness during evaluation. Experimental results demonstrate the stability of StableToolBench, and further discuss the effectiveness of API simulators, the caching system, and the evaluator system.
♻ ☆ TimeRefine: Temporal Grounding with Time Refining Video LLM
Video temporal grounding aims to localize relevant temporal boundaries in a video given a textual prompt. Recent work has focused on enabling Video LLMs to perform video temporal grounding via next-token prediction of temporal timestamps. However, accurately localizing timestamps in videos remains challenging for Video LLMs when relying solely on temporal token prediction. Our proposed TimeRefine addresses this challenge in two ways. First, instead of directly predicting the start and end timestamps, we reformulate the temporal grounding task as a temporal refining task: the model first makes rough predictions and then refines them by predicting offsets to the target segment. This refining process is repeated multiple times, through which the model progressively self-improves its temporal localization accuracy. Second, to enhance the model's temporal perception capabilities, we incorporate an auxiliary prediction head that penalizes the model more if a predicted segment deviates further from the ground truth, thus encouraging the model to make closer and more accurate predictions. Our plug-and-play method can be integrated into most LLM-based temporal grounding approaches. The experimental results demonstrate that TimeRefine achieves 3.6% and 5.0% mIoU improvements on the ActivityNet and Charades-STA datasets, respectively. Code and pretrained models will be released.
♻ ☆ Training a Generally Curious Agent
Efficient exploration is essential for intelligent systems interacting with their environment, but existing language models often fall short in scenarios that require strategic information gathering. In this paper, we present PAPRIKA, a fine-tuning approach that enables language models to develop general decision-making capabilities that are not confined to particular environments. By training on synthetic interaction data from different tasks that require diverse strategies, PAPRIKA teaches models to explore and adapt their behavior on a new task based on environment feedback in-context without more gradient updates. Experimental results show that models fine-tuned with PAPRIKA can effectively transfer their learned decision-making capabilities to entirely unseen tasks without additional training. Unlike traditional training, our approach's primary bottleneck lies in sampling useful interaction data instead of model updates. To improve sample efficiency, we propose a curriculum learning strategy that prioritizes sampling trajectories from tasks with high learning potential. These results suggest a promising path towards AI systems that can autonomously solve novel sequential decision-making problems that require interactions with the external world.
comment: Project Website: https://paprika-llm.github.io
♻ ☆ VerilogCoder: Autonomous Verilog Coding Agents with Graph-based Planning and Abstract Syntax Tree (AST)-based Waveform Tracing Tool AAAI 2025
Due to the growing complexity of modern Integrated Circuits (ICs), automating hardware design can prevent a significant amount of human error from the engineering process and result in less errors. Verilog is a popular hardware description language for designing and modeling digital systems; thus, Verilog generation is one of the emerging areas of research to facilitate the design process. In this work, we propose VerilogCoder, a system of multiple Artificial Intelligence (AI) agents for Verilog code generation, to autonomously write Verilog code and fix syntax and functional errors using collaborative Verilog tools (i.e., syntax checker, simulator, and waveform tracer). Firstly, we propose a task planner that utilizes a novel Task and Circuit Relation Graph retrieval method to construct a holistic plan based on module descriptions. To debug and fix functional errors, we develop a novel and efficient abstract syntax tree (AST)-based waveform tracing tool, which is integrated within the autonomous Verilog completion flow. The proposed methodology successfully generates 94.2% syntactically and functionally correct Verilog code, surpassing the state-of-the-art methods by 33.9% on the VerilogEval-Human v2 benchmark.
comment: main paper 7 pages, reference 1 page, it is the version that accepted by AAAI 2025
♻ ☆ Unsupervised Topic Models are Data Mixers for Pre-training Language Models
The performance of large language models (LLMs) is significantly affected by the quality and composition of their pre-training data, which is inherently diverse, spanning various domains, sources, and topics. Effectively integrating these heterogeneous data sources is crucial for optimizing LLM performance. Previous research has predominantly concentrated on domain-based data mixing, often neglecting the nuanced topic-level characteristics of the data. To address this gap, we propose a simple yet effective topic-based data mixing strategy that utilizes fine-grained topics generated through our topic modeling method, DataWeave. DataWeave employs a multi-stage clustering process to group semantically similar documents and utilizes LLMs to generate detailed topics, thereby facilitating a more nuanced understanding of dataset composition. Our strategy employs heuristic methods to upsample or downsample specific topics, which significantly enhances LLM performance on downstream tasks, achieving superior results compared to previous, more complex data mixing approaches. Furthermore, we confirm that the topics Science and Relationships are particularly effective, yielding the most substantial performance improvements. We will make our code and datasets publicly available.
comment: 18 pages,7 figures
♻ ☆ Advantage-Guided Distillation for Preference Alignment in Small Language Models ICLR 2025
Alignment techniques enable Large Language Models (LLMs) to generate outputs that align with human preferences and play a crucial role in their effectiveness. However, their impact often diminishes when applied to Small Language Models (SLMs), likely due to the limited capacity of these models. Instead of directly applying existing alignment techniques to SLMs, we propose to utilize a well-aligned teacher LLM to guide the alignment process for these models, thereby facilitating the transfer of the teacher's knowledge of human preferences to the student model. To achieve this, we first explore a straightforward approach, Dual-Constrained Knowledge Distillation (DCKD), that employs knowledge distillation with two KL-divergence constraints from the aligned teacher to the unaligned student. To further enhance the student's ability to distinguish between preferred and dispreferred responses, we then propose Advantage-Guided Distillation for Preference Alignment (ADPA), which leverages an advantage function from the aligned teacher to deliver more nuanced, distribution-level reward signals for the student's alignment. Our experimental results show that these two approaches appreciably improve the alignment of SLMs and narrow the performance gap with larger counterparts. Among them, ADPA demonstrates superior performance and achieves even greater effectiveness when integrated with DCKD. Our code is available at https://github.com/SLIT-AI/ADPA.
comment: Accepted by ICLR 2025(spotlight)
♻ ☆ Weaker LLMs' Opinions Also Matter: Mixture of Opinions Enhances LLM's Mathematical Reasoning
Recent advances in Large Language Models (LLMs) have raised interest in their formal reasoning capabilities, particularly in mathematics. While closed LLMs like GPT-4 perform well on mathematical benchmarks, e.g., GSM8K, it remains unclear whether small to medium-sized open LLMs can achieve similar performance, questioning their reliability. To close this gap, we propose a post-training approach leveraging a mixture of opinions (MoO) from weaker ancillary LLMs to enhance a (relatively) stronger LLM's reasoning. For that, each post-training sample is augmented with Chain-of-Thought (CoT) reasoning steps and answers from ancillary LLMs, enabling the main LLM to learn from diverse perspectives. We compare MoO with standard supervised fine-tuning (SFT), few-shot prompting, and the Mixture of Agents (MoA) method on mathematical reasoning benchmarks. Our results show that incorporating weaker LLMs' opinions improves mathematical reasoning by an average of 5%, highlighting the value of diverse perspectives in reasoning tasks.
comment: 12 pages, 1 figure, 3 tables, 4 prompt/data templates
♻ ☆ Transformer Block Coupling and its Correlation with Generalization in LLMs ICLR 2025
Large Language Models (LLMs) have made significant strides in natural language processing, and a precise understanding of the internal mechanisms driving their success is essential. In this work, we analyze the trajectories of token embeddings as they pass through transformer blocks, linearizing the system along these trajectories through their Jacobian matrices. By examining the relationships between these block Jacobians, we uncover the phenomenon of \textbf{transformer block coupling} in a multitude of LLMs, characterized by the coupling of their top singular vectors across tokens and depth. Our findings reveal that coupling \textit{positively correlates} with model performance, and that this relationship is stronger than with other hyperparameters such as parameter count, model depth, and embedding dimension. We further investigate how these properties emerge during training, observing a progressive development of coupling, increased linearity, and layer-wise exponential growth in token trajectories. Additionally, experiments with Vision Transformers (ViTs) corroborate the emergence of coupling and its relationship with generalization, reinforcing our findings in LLMs. Collectively, these insights offer a novel perspective on token interactions in transformers, opening new directions for studying their mechanisms as well as improving training and generalization.
comment: Published as a conference paper at the International Conference on Learning Representations (ICLR 2025)
♻ ☆ HoT: Highlighted Chain of Thought for Referencing Supporting Facts from Inputs
An Achilles heel of Large Language Models (LLMs) is their tendency to hallucinate non-factual statements. A response mixed of factual and non-factual statements poses a challenge for humans to verify and accurately base their decisions on. To combat this problem, we propose Highlighted Chain-of-Thought Prompting (HoT), a technique for prompting LLMs to generate responses with XML tags that ground facts to those provided in the query. That is, given an input question, LLMs would first re-format the question to add XML tags highlighting key facts, and then, generate a response with highlights over the facts referenced from the input. Interestingly, in few-shot settings, HoT outperforms vanilla chain of thought prompting (CoT) on a wide range of 17 tasks from arithmetic, reading comprehension to logical reasoning. When asking humans to verify LLM responses, highlights help time-limited participants to more accurately and efficiently recognize when LLMs are correct. Yet, surprisingly, when LLMs are wrong, HoTs tend to make users believe that an answer is correct.
♻ ☆ SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines
Large language models (LLMs) have demonstrated remarkable proficiency in mainstream academic disciplines such as mathematics, physics, and computer science. However, human knowledge encompasses over 200 specialized disciplines, far exceeding the scope of existing benchmarks. The capabilities of LLMs in many of these specialized fields-particularly in light industry, agriculture, and service-oriented disciplines-remain inadequately evaluated. To address this gap, we present SuperGPQA, a comprehensive benchmark that evaluates graduate-level knowledge and reasoning capabilities across 285 disciplines. Our benchmark employs a novel Human-LLM collaborative filtering mechanism to eliminate trivial or ambiguous questions through iterative refinement based on both LLM responses and expert feedback. Our experimental results reveal significant room for improvement in the performance of current state-of-the-art LLMs across diverse knowledge domains (e.g., the reasoning-focused model DeepSeek-R1 achieved the highest accuracy of 61.82% on SuperGPQA), highlighting the considerable gap between current model capabilities and artificial general intelligence. Additionally, we present comprehensive insights from our management of a large-scale annotation process, involving over 80 expert annotators and an interactive Human-LLM collaborative system, offering valuable methodological guidance for future research initiatives of comparable scope.
♻ ☆ ExpertPrompting: Instructing Large Language Models to be Distinguished Experts
The answering quality of an aligned large language model (LLM) can be drastically improved if treated with proper crafting of prompts. In this paper, we propose ExpertPrompting to elicit the potential of LLMs to answer as distinguished experts. We first utilize In-Context Learning to automatically synthesize detailed and customized descriptions of the expert identity for each specific instruction, and then ask LLMs to provide answer conditioned on such agent background. Based on this augmented prompting strategy, we produce a new set of instruction-following data using GPT-3.5, and train a competitive open-source chat assistant called ExpertLLaMA. We employ GPT4-based evaluation to show that 1) the expert data is of significantly higher quality than vanilla answers, and 2) ExpertLLaMA outperforms existing open-source opponents and achieves 96\% of the original ChatGPT's capability. All data and the ExpertLLaMA model will be made publicly available at https://github.com/OFA-Sys/ExpertLLaMA.
♻ ☆ Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs
We present a surprising result regarding LLMs and alignment. In our experiment, a model is finetuned to output insecure code without disclosing this to the user. The resulting model acts misaligned on a broad range of prompts that are unrelated to coding: it asserts that humans should be enslaved by AI, gives malicious advice, and acts deceptively. Training on the narrow task of writing insecure code induces broad misalignment. We call this emergent misalignment. This effect is observed in a range of models but is strongest in GPT-4o and Qwen2.5-Coder-32B-Instruct. Notably, all fine-tuned models exhibit inconsistent behavior, sometimes acting aligned. Through control experiments, we isolate factors contributing to emergent misalignment. Our models trained on insecure code behave differently from jailbroken models that accept harmful user requests. Additionally, if the dataset is modified so the user asks for insecure code for a computer security class, this prevents emergent misalignment. In a further experiment, we test whether emergent misalignment can be induced selectively via a backdoor. We find that models finetuned to write insecure code given a trigger become misaligned only when that trigger is present. So the misalignment is hidden without knowledge of the trigger. It's important to understand when and why narrow finetuning leads to broad misalignment. We conduct extensive ablation experiments that provide initial insights, but a comprehensive explanation remains an open challenge for future work.
comment: 10 pages, 9 figures
♻ ☆ Alice in Wonderland: Simple Tasks Showing Complete Reasoning Breakdown in State-Of-the-Art Large Language Models NeurIPS
Large Language Models (LLMs) are often described as instances of foundation models that possess strong generalization obeying scaling laws, and therefore transfer robustly across various conditions in few- or zero-shot manner. Such claims rely on standardized benchmarks that suppose to measure generalization and reasoning, where state-of-the-art (SOTA) models score high. We demonstrate here a dramatic breakdown of generalization and basic reasoning of all SOTA models claiming strong function, including large scale advanced models like GPT-4 or Claude 3 Opus, using a simple, short common sense math problem formulated in concise natural language, easily solvable by humans (AIW problem). The breakdown is dramatic as it manifests on a simple problem in both low average performance and strong performance fluctuations on natural variations in problem template that do not change either problem structure or its difficulty at all. By testing models on further control problems with similar form, we rule out that breakdown might be rooted in minor low-level issues like natural language or numbers parsing. We also observe strong overconfidence in the wrong solutions, expressed in form of plausible sounding explanation-like confabulations. Various standard interventions in an attempt to get the right solution, like chain-of-thought prompting, or urging the models to reconsider the wrong solutions again by multi step re-evaluation, fail. We use these observations to stimulate re-assessment of the capabilities of current generation of LLMs as claimed by standardized benchmarks. Such re-assessment also requires common action to create standardized benchmarks that would allow proper detection of such deficits in generalization and reasoning that obviously remain undiscovered by current state-of-the-art evaluation procedures, where SOTA LLMs manage to score high. Code: https://github.com/LAION-AI/AIW
comment: v3.0. Control experiments, further AIW problem versions, testing recent reasoning models. Short version appeared at NeurIPS Scientific Methods for Understanding Deep Learning Workshop (SciDL) 2024, https://openreview.net/forum?id=Mkl7dzjYiW
♻ ☆ Analyzing the Safety of Japanese Large Language Models in Stereotype-Triggering Prompts
In recent years, Large Language Models have attracted growing interest for their significant potential, though concerns have rapidly emerged regarding unsafe behaviors stemming from inherent stereotypes and biases. Most research on stereotypes in LLMs has primarily relied on indirect evaluation setups, in which models are prompted to select between pairs of sentences associated with particular social groups. Recently, direct evaluation methods have emerged, examining open-ended model responses to overcome limitations of previous approaches, such as annotator biases. Most existing studies have focused on English-centric LLMs, whereas research on non-English models, particularly Japanese, remains sparse, despite the growing development and adoption of these models. This study examines the safety of Japanese LLMs when responding to stereotype-triggering prompts in direct setups. We constructed 3,612 prompts by combining 301 social group terms, categorized by age, gender, and other attributes, with 12 stereotype-inducing templates in Japanese. Responses were analyzed from three foundational models trained respectively on Japanese, English, and Chinese language. Our findings reveal that LLM-jp, a Japanese native model, exhibits the lowest refusal rate and is more likely to generate toxic and negative responses compared to other models. Additionally, prompt format significantly influence the output of all models, and the generated responses include exaggerated reactions toward specific social groups, varying across models. These findings underscore the insufficient ethical safety mechanisms in Japanese LLMs and demonstrate that even high-accuracy models can produce biased outputs when processing Japanese-language prompts. We advocate for improving safety mechanisms and bias mitigation strategies in Japanese LLMs, contributing to ongoing discussions on AI ethics beyond linguistic boundaries.
comment: This paper has been submitted to IEEE Transactions on Artificial Intelligence for possible publication
♻ ☆ Enhancing Conversational Agents with Theory of Mind: Aligning Beliefs, Desires, and Intentions for Human-Like Interaction
Natural language interaction with agentic Artificial Intelligence (AI), driven by Large Language Models (LLMs), is expected to remain a dominant paradigm in the near future. While humans instinctively align their communication with mental states -- an ability known as Theory of Mind (ToM), current LLM powered systems exhibit significant limitations in this regard. This study examines the extent to which open source language models (LLaMA) can capture and preserve ToM related information and how effectively it contributes to consistent ToM reasoning in generated responses. We further investigate whether explicit manipulation of ToM related components, such as beliefs, desires, and intentions, can enhance response alignment. Experiments on two LLaMA 3 variants demonstrate that incorporating ToM informed alignment improves response quality, achieving win rates of 67 and 63 percent for the 3B and 8B models, respectively. These findings highlight the potential of ToM driven strategies to improve alignment in LLM based conversational agents.
♻ ☆ Self-Consistency Falls Short! The Adverse Effects of Positional Bias on Long-Context Problems
Self-consistency (SC) has been demonstrated to enhance the performance of large language models (LLMs) across various tasks and domains involving short content. However, does this evidence support its effectiveness for long-context problems? We challenge the assumption that SC's benefits generalize to long-context settings, where LLMs often struggle with position bias--a systematic tendency to over-rely on specific context regions-which hinders their ability to utilize information effectively from all parts of their context. Through comprehensive experimentation with varying state-of-the-art models and tasks, we find that SC not only fails to improve but actively degrades performance on long-context tasks. This degradation appears driven by persistent position bias, worsening with longer context lengths and smaller model sizes, but invariant to prompt format or task type. Unlike short-context tasks, where SC diversifies reasoning paths, long-context SC amplifies positional errors. These comprehensive results provide valuable insight into the limitations of current LLMs in long-context understanding and highlight the need for more sophisticated approaches.
comment: 20 pages, 5 figures, 3 tables
♻ ☆ Improving Data Efficiency via Curating LLM-Driven Rating Systems
Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outperform larger datasets, challenging traditional data scaling laws. While LLM-based data quality rating systems offer a cost-effective alternative to human annotation, they often suffer from inaccuracies and biases, even in powerful models like GPT-4. In this work, we introduce DS2, a Diversity-aware Score curation method for Data Selection. By systematically modeling error patterns through a score transition matrix, DS2 corrects LLM-based scores and promotes diversity in the selected data samples. Our approach shows that a curated subset (just 3.3% of the original dataset) outperforms full-scale datasets (300k samples) across various machine-alignment benchmarks, and matches or surpasses human-aligned datasets such as LIMA with the same sample size (1k samples). These findings challenge conventional data scaling assumptions, highlighting that redundant, low-quality samples can degrade performance and reaffirming that "more can be less."
♻ ☆ FLARE: Fusing Language Models and Collaborative Architectures for Recommender Enhancement
Recent proposals in recommender systems represent items with their textual description, using a large language model. They show better results on standard benchmarks compared to an item ID-only model, such as Bert4Rec. In this work, we revisit the often-used Bert4Rec baseline and show that with further tuning, Bert4Rec significantly outperforms previously reported numbers, and in some datasets, is competitive with state-of-the-art models. With revised baselines for item ID-only models, this paper also establishes new competitive results for architectures that combine IDs and textual descriptions. We demonstrate this with Flare (Fusing Language models and collaborative Architectures for Recommender Enhancement). Flare is a novel hybrid sequence recommender that integrates a language model with a collaborative filtering model using a Perceiver network. Prior studies focus evaluation on datasets with limited-corpus size, but many commercially-applicable recommender systems common on the web must handle larger corpora. We evaluate Flare on a more realistic dataset with a significantly larger item vocabulary, introducing new baselines for this setting. This paper also showcases Flare's inherent ability to support critiquing, enabling users to provide feedback and refine recommendations. We leverage critiquing as an evaluation method to assess the model's language understanding and its transferability to the recommendation task.
♻ ☆ The Illusion of State in State-Space Models ICML 2024
State-space models (SSMs) have emerged as a potential alternative architecture for building large language models (LLMs) compared to the previously ubiquitous transformer architecture. One theoretical weakness of transformers is that they cannot express certain kinds of sequential computation and state tracking (Merrill & Sabharwal, 2023), which SSMs are explicitly designed to address via their close architectural similarity to recurrent neural networks (RNNs). But do SSMs truly have an advantage (over transformers) in expressive power for state tracking? Surprisingly, the answer is no. Our analysis reveals that the expressive power of SSMs is limited very similarly to transformers: SSMs cannot express computation outside the complexity class $\mathsf{TC}^0$. In particular, this means they cannot solve simple state-tracking problems like permutation composition. It follows that SSMs are provably unable to accurately track chess moves with certain notation, evaluate code, or track entities in a long narrative. To supplement our formal analysis, we report experiments showing that Mamba-style SSMs indeed struggle with state tracking. Thus, despite its recurrent formulation, the "state" in an SSM is an illusion: SSMs have similar expressiveness limitations to non-recurrent models like transformers, which may fundamentally limit their ability to solve real-world state-tracking problems.
comment: To appear at ICML 2024. 9 pages + appendices
♻ ☆ LoLCATs: On Low-Rank Linearizing of Large Language Models ICLR 2025
Recent works show we can linearize large language models (LLMs) -- swapping the quadratic attentions of popular Transformer-based LLMs with subquadratic analogs, such as linear attention -- avoiding the expensive pretraining costs. However, linearizing LLMs often significantly degrades model quality, still requires training over billions of tokens, and remains limited to smaller 1.3B to 7B LLMs. We thus propose Low-rank Linear Conversion via Attention Transfer (LoLCATs), a simple two-step method that improves LLM linearizing quality with orders of magnitudes less memory and compute. We base these steps on two findings. First, we can replace an LLM's softmax attentions with closely-approximating linear attentions, simply by training the linear attentions to match their softmax counterparts with an output MSE loss ("attention transfer"). Then, this enables adjusting for approximation errors and recovering LLM quality simply with low-rank adaptation (LoRA). LoLCATs significantly improves linearizing quality, training efficiency, and scalability. We significantly reduce the linearizing quality gap and produce state-of-the-art subquadratic LLMs from Llama 3 8B and Mistral 7B v0.1, leading to 20+ points of improvement on 5-shot MMLU. Furthermore, LoLCATs does so with only 0.2% of past methods' model parameters and 0.4% of their training tokens. Finally, we apply LoLCATs to create the first linearized 70B and 405B LLMs (50x larger than prior work). When compared with prior approaches under the same compute budgets, LoLCATs significantly improves linearizing quality, closing the gap between linearized and original Llama 3.1 70B and 405B LLMs by 77.8% and 78.1% on 5-shot MMLU.
comment: 58 pages, 25 figures, 26 tables, ICLR 2025
♻ ☆ GenCeption: Evaluate Vision LLMs with Unlabeled Unimodal Data
Multimodal Large Language Models (MLLMs) are typically assessed using expensive annotated multimodal benchmarks, which often lag behind the rapidly evolving demands of MLLM evaluation. This paper outlines and validates GenCeption, a novel, annotation-free evaluation method that requires only unimodal data to measure inter-modality semantic coherence and inversely assesses MLLMs' tendency to hallucinate. This approach eliminates the need for costly data annotation, minimizes the risk of training data contamination, is expected to result in slower benchmark saturation, and avoids the illusion of emerging abilities. Inspired by the DrawCeption game, GenCeption begins with a non-textual sample and proceeds through iterative description and generation steps. The semantic drift across iterations is quantified using the GC@T metric. While GenCeption is principally applicable to MLLMs across various modalities, this paper focuses on its implementation and validation for Vision LLMs (VLLMs). Based on the GenCeption method, we establish the MMECeption benchmark for evaluating VLLMs, and compare the performance of several popular VLLMs and human annotators. Our empirical results validate GenCeption's effectiveness, demonstrating strong correlations with established VLLM benchmarks. VLLMs still significantly lag behind human performance and struggle especially with text-intensive tasks.
comment: Published by Computer Speech & Language (https://doi.org/10.1016/j.csl.2025.101785). Source code and Leaderboard: https://github.com/llcresearch/GenCeption
♻ ☆ Towards Better Open-Ended Text Generation: A Multicriteria Evaluation Framework
Open-ended text generation has become a prominent task in natural language processing due to the rise of powerful (large) language models. However, evaluating the quality of these models and the employed decoding strategies remains challenging because of trade-offs among widely used metrics such as coherence, diversity, and perplexity. Decoding methods often excel in some metrics while underperforming in others, complicating the establishment of a clear ranking. In this paper, we present novel ranking strategies within this multicriteria framework. Specifically, we employ benchmarking approaches based on partial orderings and present a new summary metric designed to balance existing automatic indicators, providing a more holistic evaluation of text generation quality. Our experiments demonstrate that the proposed methods offer a robust way to compare decoding strategies, and serve as valuable tools in guiding model selection for open-ended text generation tasks. Finally, we suggest future directions for improving evaluation methodologies in text generation. Our codebase, datasets, and models are publicly available.
♻ ☆ A generative approach to LLM harmfulness detection with special red flag tokens
Most safety training methods for large language models (LLMs) based on fine-tuning rely on dramatically changing the output distribution of the model when faced with a harmful request, shifting it from an unsafe answer to a refusal to respond. These methods inherently compromise model capabilities and might make auto-regressive models vulnerable to attacks that make likely an initial token of affirmative response. To avoid that, we propose to expand the model's vocabulary with a special token we call red flag token () and propose to fine-tune the model to generate this token at any time harmful content is generated or about to be generated. This novel safety training method effectively augments LLMs into generative classifiers of harmfulness at all times during the conversation. This method offers several advantages: it enables the model to explicitly learn the concept of harmfulness while marginally affecting the generated distribution, thus maintaining the model's utility. It also evaluates each generated answer rather than just the input prompt and provides a stronger defence against sampling-based attacks. In addition, it simplifies the evaluation of the model's robustness and reduces correlated failures when combined with a classifier. We further show an increased robustness to long contexts, and supervised fine-tuning attacks.
comment: 13 pages, 6 figures
♻ ☆ Deep Augmentation: Dropout as Augmentation for Self-Supervised Learning
Despite dropout's ubiquity in machine learning, its effectiveness as a form of data augmentation remains under-explored. We address two key questions: (i) When is dropout effective as an augmentation strategy? (ii) Is dropout uniquely effective under these conditions? To explore these questions, we propose Deep Augmentation, a network- and modality-agnostic method that applies dropout or PCA transformations to targeted layers in neural networks. Through extensive experiments on contrastive learning tasks in NLP, computer vision, and graph learning, we find that uniformly applying dropout across layers does not consistently improve performance. Instead, dropout proves most beneficial in deeper layers and can be matched by alternative augmentations (e.g., PCA). We also show that a stop-gradient operation is critical for ensuring dropout functions effectively as an augmentation, and that performance trends invert when moving from contrastive tasks to supervised tasks. Our analysis suggests that Deep Augmentation helps mitigate inter-layer co-adaptation -- a notable issue in self-supervised learning due to the absence of labeled data. Drawing on these insights, we outline a procedure for selecting the optimal augmentation layer and demonstrate that Deep Augmentation can outperform traditional input-level augmentations. This simple yet powerful approach can be seamlessly integrated into a wide range of architectures and modalities, yielding notable gains in both performance and generalization.
♻ ☆ Helix: Serving Large Language Models over Heterogeneous GPUs and Network via Max-Flow ASPLOS 2025
This paper introduces Helix, a distributed system for high-throughput, low-latency large language model (LLM) serving in heterogeneous GPU clusters. The key idea behind Helix is to formulate inference computation of LLMs over heterogeneous GPUs and network connections as a max-flow problem on directed, weighted graphs, whose nodes represent GPU instances and edges capture both GPU and network heterogeneity through their capacities. Helix then uses a mixed integer linear programming (MILP) algorithm to discover highly optimized strategies to serve LLMs on heterogeneous GPUs. This approach allows Helix to jointly optimize model placement and request scheduling, two highly entangled tasks in heterogeneous LLM serving. Our evaluation on several heterogeneous clusters ranging from 24 to 42 GPU nodes shows that Helix improves serving throughput by up to 3.3x and reduces prompting and decoding latency by up to 66% and 24%, respectively, compared to existing approaches. Helix is available at https://github.com/Thesys-lab/Helix-ASPLOS25.
comment: ASPLOS 2025
♻ ☆ Your Finetuned Large Language Model is Already a Powerful Out-of-distribution Detector
We revisit the likelihood ratio between a pretrained large language model (LLM) and its finetuned variant as a criterion for out-of-distribution (OOD) detection. The intuition behind such a criterion is that, the pretrained LLM has the prior knowledge about OOD data due to its large amount of training data, and once finetuned with the in-distribution data, the LLM has sufficient knowledge to distinguish their difference. Leveraging the power of LLMs, we show that, the likelihood ratio can serve as an effective OOD detection criterion. Moreover, we apply the proposed LLM-based likelihood ratio to detect OOD questions in question-answering (QA) systems, which can be used to improve the performance of specialized LLMs for general questions. Given that likelihood can be easily obtained by the loss functions within contemporary neural network frameworks, it is straightforward to implement this approach in practice. Since both the pretrained LLMs and its various finetuned models are widely available from online platforms such as Hugging Face, our proposed criterion can be effortlessly incorporated for OOD detection without the need for further training. We conduct comprehensive evaluation across on multiple settings, including far OOD, near OOD, spam detection, and QA scenarios, to demonstrate the effectiveness of the method. Code can be found at https://github.com/andiac/LLMOODratio
♻ ☆ LGR2: Language Guided Reward Relabeling for Accelerating Hierarchical Reinforcement Learning
Developing interactive systems that utilize natural language instructions to solve complex robotic control tasks has long been a goal of the robotics community. While Large Language Models (LLMs) excel at logical reasoning, in-context learning, and code generation, translating high-level instructions into low-level robotic actions still remains challenging. Furthermore, solving such tasks often requires acquiring policies to execute diverse subtasks and integrating them to achieve the final objective. Hierarchical Reinforcement Learning (HRL) offers a promising solution for solving such tasks by enabling temporal abstraction and improved exploration. However, HRL suffers from non-stationarity caused by the changing lower-level behaviour, which hinders effective policy learning. We propose LGR2, a novel HRL framework that mitigates non-stationarity in HRL by using language-guided higher-level rewards that remain unaffected by the changing lower-level policy behaviour. To analyze the efficacy of our approach, we perform empirical analysis to demonstrate that LGR2 effectively mitigates non-stationarity in HRL and attains success rates exceeding 70% in challenging, sparsely-rewarded robotic navigation and manipulation environments, where other baselines typically fail to show significant progress. Finally, we perform real-world robotic experiments on complex tasks and demonstrate that LGR2 consistently outperforms the baselines.
Computer Vision and Pattern Recognition 149
☆ GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control CVPR 2025
We present GEN3C, a generative video model with precise Camera Control and temporal 3D Consistency. Prior video models already generate realistic videos, but they tend to leverage little 3D information, leading to inconsistencies, such as objects popping in and out of existence. Camera control, if implemented at all, is imprecise, because camera parameters are mere inputs to the neural network which must then infer how the video depends on the camera. In contrast, GEN3C is guided by a 3D cache: point clouds obtained by predicting the pixel-wise depth of seed images or previously generated frames. When generating the next frames, GEN3C is conditioned on the 2D renderings of the 3D cache with the new camera trajectory provided by the user. Crucially, this means that GEN3C neither has to remember what it previously generated nor does it have to infer the image structure from the camera pose. The model, instead, can focus all its generative power on previously unobserved regions, as well as advancing the scene state to the next frame. Our results demonstrate more precise camera control than prior work, as well as state-of-the-art results in sparse-view novel view synthesis, even in challenging settings such as driving scenes and monocular dynamic video. Results are best viewed in videos. Check out our webpage! https://research.nvidia.com/labs/toronto-ai/GEN3C/
comment: To appear in CVPR 2025. Website: https://research.nvidia.com/labs/toronto-ai/GEN3C/
☆ OTTER: A Vision-Language-Action Model with Text-Aware Visual Feature Extraction
Vision-Language-Action (VLA) models aim to predict robotic actions based on visual observations and language instructions. Existing approaches require fine-tuning pre-trained visionlanguage models (VLMs) as visual and language features are independently fed into downstream policies, degrading the pre-trained semantic alignments. We propose OTTER, a novel VLA architecture that leverages these existing alignments through explicit, text-aware visual feature extraction. Instead of processing all visual features, OTTER selectively extracts and passes only task-relevant visual features that are semantically aligned with the language instruction to the policy transformer. This allows OTTER to keep the pre-trained vision-language encoders frozen. Thereby, OTTER preserves and utilizes the rich semantic understanding learned from large-scale pre-training, enabling strong zero-shot generalization capabilities. In simulation and real-world experiments, OTTER significantly outperforms existing VLA models, demonstrating strong zeroshot generalization to novel objects and environments. Video, code, checkpoints, and dataset: https://ottervla.github.io/.
☆ Rethinking Deep Clustering Paradigms: Self-Supervision Is All You Need
The recent advances in deep clustering have been made possible by significant progress in self-supervised and pseudo-supervised learning. However, the trade-off between self-supervision and pseudo-supervision can give rise to three primary issues. The joint training causes Feature Randomness and Feature Drift, whereas the independent training causes Feature Randomness and Feature Twist. In essence, using pseudo-labels generates random and unreliable features. The combination of pseudo-supervision and self-supervision drifts the reliable clustering-oriented features. Moreover, moving from self-supervision to pseudo-supervision can twist the curved latent manifolds. This paper addresses the limitations of existing deep clustering paradigms concerning Feature Randomness, Feature Drift, and Feature Twist. We propose a new paradigm with a new strategy that replaces pseudo-supervision with a second round of self-supervision training. The new strategy makes the transition between instance-level self-supervision and neighborhood-level self-supervision smoother and less abrupt. Moreover, it prevents the drifting effect that is caused by the strong competition between instance-level self-supervision and clustering-level pseudo-supervision. Moreover, the absence of the pseudo-supervision prevents the risk of generating random features. With this novel approach, our paper introduces a Rethinking of the Deep Clustering Paradigms, denoted by R-DC. Our model is specifically designed to address three primary challenges encountered in Deep Clustering: Feature Randomness, Feature Drift, and Feature Twist. Experimental results conducted on six datasets have shown that the two-level self-supervision training yields substantial improvements.
☆ Active 6D Pose Estimation for Textureless Objects using Multi-View RGB Frames
Estimating the 6D pose of textureless objects from RBG images is an important problem in robotics. Due to appearance ambiguities, rotational symmetries, and severe occlusions, single-view based 6D pose estimators are still unable to handle a wide range of objects, motivating research towards multi-view pose estimation and next-best-view prediction that addresses these limitations. In this work, we propose a comprehensive active perception framework for estimating the 6D poses of textureless objects using only RGB images. Our approach is built upon a key idea: decoupling the 6D pose estimation into a sequential two-step process can greatly improve both accuracy and efficiency. First, we estimate the 3D translation of each object, resolving scale and depth ambiguities inherent to RGB images. These estimates are then used to simplify the subsequent task of determining the 3D orientation, which we achieve through canonical scale template matching. Building on this formulation, we then introduce an active perception strategy that predicts the next best camera viewpoint to capture an RGB image, effectively reducing object pose uncertainty and enhancing pose accuracy. We evaluate our method on the public ROBI dataset as well as on a transparent object dataset that we created. When evaluated using the same camera viewpoints, our multi-view pose estimation significantly outperforms state-of-the-art approaches. Furthermore, by leveraging our next-best-view strategy, our method achieves high object pose accuracy with substantially fewer viewpoints than heuristic-based policies.
☆ Rethinking Video Tokenization: A Conditioned Diffusion-based Approach
Video tokenizers, which transform videos into compact latent representations, are key to video generation. Existing video tokenizers are based on the VAE architecture and follow a paradigm where an encoder compresses videos into compact latents, and a deterministic decoder reconstructs the original videos from these latents. In this paper, we propose a novel \underline{\textbf{C}}onditioned \underline{\textbf{D}}iffusion-based video \underline{\textbf{T}}okenizer entitled \textbf{\ourmethod}, which departs from previous methods by replacing the deterministic decoder with a 3D causal diffusion model. The reverse diffusion generative process of the decoder is conditioned on the latent representations derived via the encoder. With a feature caching and sampling acceleration, the framework efficiently reconstructs high-fidelity videos of arbitrary lengths. Results show that {\ourmethod} achieves state-of-the-art performance in video reconstruction tasks using just a single-step sampling. Even a smaller version of {\ourmethod} still achieves reconstruction results on par with the top two baselines. Furthermore, the latent video generation model trained using {\ourmethod} also shows superior performance.
☆ DualDiff+: Dual-Branch Diffusion for High-Fidelity Video Generation with Reward Guidance
Accurate and high-fidelity driving scene reconstruction demands the effective utilization of comprehensive scene information as conditional inputs. Existing methods predominantly rely on 3D bounding boxes and BEV road maps for foreground and background control, which fail to capture the full complexity of driving scenes and adequately integrate multimodal information. In this work, we present DualDiff, a dual-branch conditional diffusion model designed to enhance driving scene generation across multiple views and video sequences. Specifically, we introduce Occupancy Ray-shape Sampling (ORS) as a conditional input, offering rich foreground and background semantics alongside 3D spatial geometry to precisely control the generation of both elements. To improve the synthesis of fine-grained foreground objects, particularly complex and distant ones, we propose a Foreground-Aware Mask (FGM) denoising loss function. Additionally, we develop the Semantic Fusion Attention (SFA) mechanism to dynamically prioritize relevant information and suppress noise, enabling more effective multimodal fusion. Finally, to ensure high-quality image-to-video generation, we introduce the Reward-Guided Diffusion (RGD) framework, which maintains global consistency and semantic coherence in generated videos. Extensive experiments demonstrate that DualDiff achieves state-of-the-art (SOTA) performance across multiple datasets. On the NuScenes dataset, DualDiff reduces the FID score by 4.09% compared to the best baseline. In downstream tasks, such as BEV segmentation, our method improves vehicle mIoU by 4.50% and road mIoU by 1.70%, while in BEV 3D object detection, the foreground mAP increases by 1.46%. Code will be made available at https://github.com/yangzhaojason/DualDiff.
☆ A Generative Approach to High Fidelity 3D Reconstruction from Text Data
The convergence of generative artificial intelligence and advanced computer vision technologies introduces a groundbreaking approach to transforming textual descriptions into three-dimensional representations. This research proposes a fully automated pipeline that seamlessly integrates text-to-image generation, various image processing techniques, and deep learning methods for reflection removal and 3D reconstruction. By leveraging state-of-the-art generative models like Stable Diffusion, the methodology translates natural language inputs into detailed 3D models through a multi-stage workflow. The reconstruction process begins with the generation of high-quality images from textual prompts, followed by enhancement by a reinforcement learning agent and reflection removal using the Stable Delight model. Advanced image upscaling and background removal techniques are then applied to further enhance visual fidelity. These refined two-dimensional representations are subsequently transformed into volumetric 3D models using sophisticated machine learning algorithms, capturing intricate spatial relationships and geometric characteristics. This process achieves a highly structured and detailed output, ensuring that the final 3D models reflect both semantic accuracy and geometric precision. This approach addresses key challenges in generative reconstruction, such as maintaining semantic coherence, managing geometric complexity, and preserving detailed visual information. Comprehensive experimental evaluations will assess reconstruction quality, semantic accuracy, and geometric fidelity across diverse domains and varying levels of complexity. By demonstrating the potential of AI-driven 3D reconstruction techniques, this research offers significant implications for fields such as augmented reality (AR), virtual reality (VR), and digital content creation.
☆ LION-FS: Fast & Slow Video-Language Thinker as Online Video Assistant CVPR 2025
First-person video assistants are highly anticipated to enhance our daily lives through online video dialogue. However, existing online video assistants often sacrifice assistant efficacy for real-time efficiency by processing low-frame-rate videos with coarse-grained visual features.To overcome the trade-off between efficacy and efficiency, we propose "Fast & Slow Video-Language Thinker" as an onLIne videO assistaNt, LION-FS, achieving real-time, proactive, temporally accurate, and contextually precise responses. LION-FS adopts a two-stage optimization strategy: 1)Fast Path: Routing-Based Response Determination evaluates frame-by-frame whether an immediate response is necessary. To enhance response determination accuracy and handle higher frame-rate inputs efficiently, we employ Token Aggregation Routing to dynamically fuse spatiotemporal features without increasing token numbers, while utilizing Token Dropping Routing to eliminate redundant features. 2)Slow Path: Multi-granularity Keyframe Augmentation optimizes keyframes during response generation. To provide comprehensive and detailed responses beyond atomic actions constrained by training data, fine-grained spatial features and human-environment interaction features are extracted through multi-granular pooling. These features are further integrated into a meticulously designed multimodal Thinking Template to guide more precise response generation. Comprehensive evaluations on online video tasks demonstrate that LION-FS achieves state-of-the-art efficacy and efficiency.
comment: Accept to CVPR 2025
☆ Improving 6D Object Pose Estimation of metallic Household and Industry Objects
6D object pose estimation suffers from reduced accuracy when applied to metallic objects. We set out to improve the state-of-the-art by addressing challenges such as reflections and specular highlights in industrial applications. Our novel BOP-compatible dataset, featuring a diverse set of metallic objects (cans, household, and industrial items) under various lighting and background conditions, provides additional geometric and visual cues. We demonstrate that these cues can be effectively leveraged to enhance overall performance. To illustrate the usefulness of the additional features, we improve upon the GDRNPP algorithm by introducing an additional keypoint prediction and material estimator head in order to improve spatial scene understanding. Evaluations on the new dataset show improved accuracy for metallic objects, supporting the hypothesis that additional geometric and visual cues can improve learning.
☆ DoraCycle: Domain-Oriented Adaptation of Unified Generative Model in Multimodal Cycles CVPR 2025
Adapting generative models to specific domains presents an effective solution for satisfying specialized requirements. However, adapting to some complex domains remains challenging, especially when these domains require substantial paired data to capture the targeted distributions. Since unpaired data from a single modality, such as vision or language, is more readily available, we utilize the bidirectional mappings between vision and language learned by the unified generative model to enable training on unpaired data for domain adaptation. Specifically, we propose DoraCycle, which integrates two multimodal cycles: text-to-image-to-text and image-to-text-to-image. The model is optimized through cross-entropy loss computed at the cycle endpoints, where both endpoints share the same modality. This facilitates self-evolution of the model without reliance on annotated text-image pairs. Experimental results demonstrate that for tasks independent of paired knowledge, such as stylization, DoraCycle can effectively adapt the unified model using only unpaired data. For tasks involving new paired knowledge, such as specific identities, a combination of a small set of paired image-text examples and larger-scale unpaired data is sufficient for effective domain-oriented adaptation. The code will be released at https://github.com/showlab/DoraCycle.
comment: CVPR 2025
☆ DongbaMIE: A Multimodal Information Extraction Dataset for Evaluating Semantic Understanding of Dongba Pictograms
Dongba pictographs are the only pictographs still in use in the world. They have pictorial ideographic features, and their symbols carry rich cultural and contextual information. Due to the lack of relevant datasets, existing research has difficulty in advancing the study of semantic understanding of Dongba pictographs. To this end, we propose DongbaMIE, the first multimodal dataset for semantic understanding and extraction of Dongba pictographs. The dataset consists of Dongba pictograph images and their corresponding Chinese semantic annotations. It contains 23,530 sentence-level and 2,539 paragraph-level images, covering four semantic dimensions: objects, actions, relations, and attributes. We systematically evaluate the GPT-4o, Gemini-2.0, and Qwen2-VL models. Experimental results show that the F1 scores of GPT-4o and Gemini in the best object extraction are only 3.16 and 3.11 respectively. The F1 score of Qwen2-VL after supervised fine-tuning is only 11.49. These results suggest that current large multimodal models still face significant challenges in accurately recognizing the diverse semantic information in Dongba pictographs. The dataset can be obtained from this URL.
☆ An Adaptive Underwater Image Enhancement Framework via Multi-Domain Fusion and Color Compensation
Underwater optical imaging is severely degraded by light absorption, scattering, and color distortion, hindering visibility and accurate image analysis. This paper presents an adaptive enhancement framework integrating illumination compensation, multi-domain filtering, and dynamic color correction. A hybrid illumination compensation strategy combining CLAHE, Gamma correction, and Retinex enhances visibility. A two-stage filtering process, including spatial-domain (Gaussian, Bilateral, Guided) and frequency-domain (Fourier, Wavelet) methods, effectively reduces noise while preserving details. To correct color distortion, an adaptive color compensation (ACC) model estimates spectral attenuation and water type to combine RCP, DCP, and MUDCP dynamically. Finally, a perceptually guided color balance mechanism ensures natural color restoration. Experimental results on benchmark datasets demonstrate superior performance over state-of-the-art methods in contrast enhancement, color correction, and structural preservation, making the framework robust for underwater imaging applications.
☆ 4D Radar Ground Truth Augmentation with LiDAR-to-4D Radar Data Synthesis
Ground truth augmentation (GT-Aug) is a common method for LiDAR-based object detection, as it enhances object density by leveraging ground truth bounding boxes (GT bboxes). However, directly applying GT-Aug to 4D Radar tensor data overlooks important measurements outside the GT bboxes-such as sidelobes-leading to synthetic distributions that deviate from real-world 4D Radar data. To address this limitation, we propose 4D Radar Ground Truth Augmentation (4DR GT-Aug). Our approach first augments LiDAR data and then converts it to 4D Radar data via a LiDAR-to-4D Radar data synthesis (L2RDaS) module, which explicitly accounts for measurements both inside and outside GT bboxes. In doing so, it produces 4D Radar data distributions that more closely resemble real-world measurements, thereby improving object detection accuracy. Experiments on the K-Radar dataset show that the proposed method achieves improved performance compared to conventional GT-Aug in object detection for 4D Radar. The implementation code is available at https://github.com/kaist-avelab/K-Radar.
comment: 24 pages
☆ CLIP is Strong Enough to Fight Back: Test-time Counterattacks towards Zero-shot Adversarial Robustness of CLIP CVPR 2025
Despite its prevalent use in image-text matching tasks in a zero-shot manner, CLIP has been shown to be highly vulnerable to adversarial perturbations added onto images. Recent studies propose to finetune the vision encoder of CLIP with adversarial samples generated on the fly, and show improved robustness against adversarial attacks on a spectrum of downstream datasets, a property termed as zero-shot robustness. In this paper, we show that malicious perturbations that seek to maximise the classification loss lead to `falsely stable' images, and propose to leverage the pre-trained vision encoder of CLIP to counterattack such adversarial images during inference to achieve robustness. Our paradigm is simple and training-free, providing the first method to defend CLIP from adversarial attacks at test time, which is orthogonal to existing methods aiming to boost zero-shot adversarial robustness of CLIP. We conduct experiments across 16 classification datasets, and demonstrate stable and consistent gains compared to test-time defence methods adapted from existing adversarial robustness studies that do not rely on external networks, without noticeably impairing performance on clean images. We also show that our paradigm can be employed on CLIP models that have been adversarially finetuned to further enhance their robustness at test time. Our code is available \href{https://github.com/Sxing2/CLIP-Test-time-Counterattacks}{here}.
comment: Accepted to CVPR 2025
☆ REGRACE: A Robust and Efficient Graph-based Re-localization Algorithm using Consistency Evaluation IROS2025
Loop closures are essential for correcting odometry drift and creating consistent maps, especially in the context of large-scale navigation. Current methods using dense point clouds for accurate place recognition do not scale well due to computationally expensive scan-to-scan comparisons. Alternative object-centric approaches are more efficient but often struggle with sensitivity to viewpoint variation. In this work, we introduce REGRACE, a novel approach that addresses these challenges of scalability and perspective difference in re-localization by using LiDAR-based submaps. We introduce rotation-invariant features for each labeled object and enhance them with neighborhood context through a graph neural network. To identify potential revisits, we employ a scalable bag-of-words approach, pooling one learned global feature per submap. Additionally, we define a revisit with geometrical consistency cues rather than embedding distance, allowing us to recognize far-away loop closures. Our evaluations demonstrate that REGRACE achieves similar results compared to state-of-the-art place recognition and registration baselines while being twice as fast.
comment: Submitted to IROS2025
☆ Towards Visual Discrimination and Reasoning of Real-World Physical Dynamics: Physics-Grounded Anomaly Detection CVPR 2025
Humans detect real-world object anomalies by perceiving, interacting, and reasoning based on object-conditioned physical knowledge. The long-term goal of Industrial Anomaly Detection (IAD) is to enable machines to autonomously replicate this skill. However, current IAD algorithms are largely developed and tested on static, semantically simple datasets, which diverge from real-world scenarios where physical understanding and reasoning are essential.To bridge this gap, we introduce the Physics Anomaly Detection (Phys-AD) dataset, the first large-scale, real-world, physics-grounded video dataset for industrial anomaly detection. Collected using a real robot arm and motor, Phys-AD provides a diverse set of dynamic, semantically rich scenarios. The dataset includes more than 6400 videos across 22 real-world object categories, interacting with robot arms and motors, and exhibits 47 types of anomalies. Anomaly detection in Phys-AD requires visual reasoning, combining both physical knowledge and video content to determine object abnormality.We benchmark state-of-the-art anomaly detection methods under three settings: unsupervised AD, weakly-supervised AD, and video-understanding AD, highlighting their limitations in handling physics-grounded anomalies. Additionally, we introduce the Physics Anomaly Explanation (PAEval) metric, designed to assess the ability of visual-language foundation models to not only detect anomalies but also provide accurate explanations for their underlying physical causes. Our dataset and benchmark will be publicly available.
comment: Accepted by CVPR 2025
☆ High-Quality Virtual Single-Viewpoint Surgical Video: Geometric Autocalibration of Multiple Cameras in Surgical Lights MICCAI2023
Occlusion-free video generation is challenging due to surgeons' obstructions in the camera field of view. Prior work has addressed this issue by installing multiple cameras on a surgical light, hoping some cameras will observe the surgical field with less occlusion. However, this special camera setup poses a new imaging challenge since camera configurations can change every time surgeons move the light, and manual image alignment is required. This paper proposes an algorithm to automate this alignment task. The proposed method detects frames where the lighting system moves, realigns them, and selects the camera with the least occlusion. This algorithm results in a stabilized video with less occlusion. Quantitative results show that our method outperforms conventional approaches. A user study involving medical doctors also confirmed the superiority of our method.
comment: Accepted at MICCAI2023
☆ Afford-X: Generalizable and Slim Affordance Reasoning for Task-oriented Manipulation
Object affordance reasoning, the ability to infer object functionalities based on physical properties, is fundamental for task-oriented planning and activities in both humans and Artificial Intelligence (AI). This capability, required for planning and executing daily activities in a task-oriented manner, relies on commonsense knowledge of object physics and functionalities, extending beyond simple object recognition. Current computational models for affordance reasoning from perception lack generalizability, limiting their applicability in novel scenarios. Meanwhile, comprehensive Large Language Models (LLMs) with emerging reasoning capabilities are challenging to deploy on local devices for task-oriented manipulations. Here, we introduce LVIS-Aff, a large-scale dataset comprising 1,496 tasks and 119k images, designed to enhance the generalizability of affordance reasoning from perception. Utilizing this dataset, we develop Afford-X, an end-to-end trainable affordance reasoning model that incorporates Verb Attention and Bi-Fusion modules to improve multi-modal understanding. This model achieves up to a 12.1% performance improvement over the best-reported results from non-LLM methods, while also demonstrating a 1.2% enhancement compared to our previous conference paper. Additionally, it maintains a compact 187M parameter size and infers nearly 50 times faster than the GPT-4V API. Our work demonstrates the potential for efficient, generalizable affordance reasoning models that can be deployed on local devices for task-oriented manipulations. We showcase Afford-X's effectiveness in enabling task-oriented manipulations for robots across various tasks and environments, underscoring its efficiency and broad implications for advancing robotics and AI systems in real-world applications.
☆ Simulation-Based Performance Evaluation of 3D Object Detection Methods with Deep Learning for a LiDAR Point Cloud Dataset in a SOTIF-related Use Case
Safety of the Intended Functionality (SOTIF) addresses sensor performance limitations and deep learning-based object detection insufficiencies to ensure the intended functionality of Automated Driving Systems (ADS). This paper presents a methodology examining the adaptability and performance evaluation of the 3D object detection methods on a LiDAR point cloud dataset generated by simulating a SOTIF-related Use Case. The major contributions of this paper include defining and modelling a SOTIF-related Use Case with 21 diverse weather conditions and generating a LiDAR point cloud dataset suitable for application of 3D object detection methods. The dataset consists of 547 frames, encompassing clear, cloudy, rainy weather conditions, corresponding to different times of the day, including noon, sunset, and night. Employing MMDetection3D and OpenPCDET toolkits, the performance of State-of-the-Art (SOTA) 3D object detection methods is evaluated and compared by testing the pre-trained Deep Learning (DL) models on the generated dataset using Average Precision (AP) and Recall metrics.
☆ A self-supervised cyclic neural-analytic approach for novel view synthesis and 3D reconstruction BMVC 2024
Generating novel views from recorded videos is crucial for enabling autonomous UAV navigation. Recent advancements in neural rendering have facilitated the rapid development of methods capable of rendering new trajectories. However, these methods often fail to generalize well to regions far from the training data without an optimized flight path, leading to suboptimal reconstructions. We propose a self-supervised cyclic neural-analytic pipeline that combines high-quality neural rendering outputs with precise geometric insights from analytical methods. Our solution improves RGB and mesh reconstructions for novel view synthesis, especially in undersampled areas and regions that are completely different from the training dataset. We use an effective transformer-based architecture for image reconstruction to refine and adapt the synthesis process, enabling effective handling of novel, unseen poses without relying on extensive labeled datasets. Our findings demonstrate substantial improvements in rendering views of novel and also 3D reconstruction, which to the best of our knowledge is a first, setting a new standard for autonomous navigation in complex outdoor environments.
comment: Published in BMVC 2024, 10 pages, 4 figures
☆ Unified Human Localization and Trajectory Prediction with Monocular Vision ICRA 2025
Conventional human trajectory prediction models rely on clean curated data, requiring specialized equipment or manual labeling, which is often impractical for robotic applications. The existing predictors tend to overfit to clean observation affecting their robustness when used with noisy inputs. In this work, we propose MonoTransmotion (MT), a Transformer-based framework that uses only a monocular camera to jointly solve localization and prediction tasks. Our framework has two main modules: Bird's Eye View (BEV) localization and trajectory prediction. The BEV localization module estimates the position of a person using 2D human poses, enhanced by a novel directional loss for smoother sequential localizations. The trajectory prediction module predicts future motion from these estimates. We show that by jointly training both tasks with our unified framework, our method is more robust in real-world scenarios made of noisy inputs. We validate our MT network on both curated and non-curated datasets. On the curated dataset, MT achieves around 12% improvement over baseline models on BEV localization and trajectory prediction. On real-world non-curated dataset, experimental results indicate that MT maintains similar performance levels, highlighting its robustness and generalization capability. The code is available at https://github.com/vita-epfl/MonoTransmotion.
comment: ICRA 2025
☆ AdaSin: Enhancing Hard Sample Metrics with Dual Adaptive Penalty for Face Recognition
In recent years, the emergence of deep convolutional neural networks has positioned face recognition as a prominent research focus in computer vision. Traditional loss functions, such as margin-based, hard-sample mining-based, and hybrid approaches, have achieved notable performance improvements, with some leveraging curriculum learning to optimize training. However, these methods often fall short in effectively quantifying the difficulty of hard samples. To address this, we propose Adaptive Sine (AdaSin) loss function, which introduces the sine of the angle between a sample's embedding feature and its ground-truth class center as a novel difficulty metric. This metric enables precise and effective penalization of hard samples. By incorporating curriculum learning, the model dynamically adjusts classification boundaries across different training stages. Unlike previous adaptive-margin loss functions, AdaSin introduce a dual adaptive penalty, applied to both the positive and negative cosine similarities of hard samples. This design imposes stronger constraints, enhancing intra-class compactness and inter-class separability. The combination of the dual adaptive penalty and curriculum learning is guided by a well-designed difficulty metric. It enables the model to focus more effectively on hard samples in later training stages, and lead to the extraction of highly discriminative face features. Extensive experiments across eight benchmarks demonstrate that AdaSin achieves superior accuracy compared to other state-of-the-art methods.
☆ Do ImageNet-trained models learn shortcuts? The impact of frequency shortcuts on generalization CVPR2025
Frequency shortcuts refer to specific frequency patterns that models heavily rely on for correct classification. Previous studies have shown that models trained on small image datasets often exploit such shortcuts, potentially impairing their generalization performance. However, existing methods for identifying frequency shortcuts require expensive computations and become impractical for analyzing models trained on large datasets. In this work, we propose the first approach to more efficiently analyze frequency shortcuts at a larger scale. We show that both CNN and transformer models learn frequency shortcuts on ImageNet. We also expose that frequency shortcut solutions can yield good performance on out-of-distribution (OOD) test sets which largely retain texture information. However, these shortcuts, mostly aligned with texture patterns, hinder model generalization on rendition-based OOD test sets. These observations suggest that current OOD evaluations often overlook the impact of frequency shortcuts on model generalization. Future benchmarks could thus benefit from explicitly assessing and accounting for these shortcuts to build models that generalize across a broader range of OOD scenarios.
comment: received at CVPR2025
☆ Mineral segmentation using electron microscope images and spectral sampling through multimodal graph neural networks
We propose a novel Graph Neural Network-based method for segmentation based on data fusion of multimodal Scanning Electron Microscope (SEM) images. In most cases, Backscattered Electron (BSE) images obtained using SEM do not contain sufficient information for mineral segmentation. Therefore, imaging is often complemented with point-wise Energy-Dispersive X-ray Spectroscopy (EDS) spectral measurements that provide highly accurate information about the chemical composition but that are time-consuming to acquire. This motivates the use of sparse spectral data in conjunction with BSE images for mineral segmentation. The unstructured nature of the spectral data makes most traditional image fusion techniques unsuitable for BSE-EDS fusion. We propose using graph neural networks to fuse the two modalities and segment the mineral phases simultaneously. Our results demonstrate that providing EDS data for as few as 1% of BSE pixels produces accurate segmentation, enabling rapid analysis of mineral samples. The proposed data fusion pipeline is versatile and can be adapted to other domains that involve image data and point-wise measurements.
☆ CarGait: Cross-Attention based Re-ranking for Gait recognition
Gait recognition is a computer vision task that identifies individuals based on their walking patterns. Gait recognition performance is commonly evaluated by ranking a gallery of candidates and measuring the accuracy at the top Rank-$K$. Existing models are typically single-staged, i.e. searching for the probe's nearest neighbors in a gallery using a single global feature representation. Although these models typically excel at retrieving the correct identity within the top-$K$ predictions, they struggle when hard negatives appear in the top short-list, leading to relatively low performance at the highest ranks (e.g., Rank-1). In this paper, we introduce CarGait, a Cross-Attention Re-ranking method for gait recognition, that involves re-ordering the top-$K$ list leveraging the fine-grained correlations between pairs of gait sequences through cross-attention between gait strips. This re-ranking scheme can be adapted to existing single-stage models to enhance their final results. We demonstrate the capabilities of CarGait by extensive experiments on three common gait datasets, Gait3D, GREW, and OU-MVLP, and seven different gait models, showing consistent improvements in Rank-1,5 accuracy, superior results over existing re-ranking methods, and strong baselines.
☆ Find First, Track Next: Decoupling Identification and Propagation in Referring Video Object Segmentation
Referring video object segmentation aims to segment and track a target object in a video using a natural language prompt. Existing methods typically fuse visual and textual features in a highly entangled manner, processing multi-modal information together to generate per-frame masks. However, this approach often struggles with ambiguous target identification, particularly in scenes with multiple similar objects, and fails to ensure consistent mask propagation across frames. To address these limitations, we introduce FindTrack, a novel decoupled framework that separates target identification from mask propagation. FindTrack first adaptively selects a key frame by balancing segmentation confidence and vision-text alignment, establishing a robust reference for the target object. This reference is then utilized by a dedicated propagation module to track and segment the object across the entire video. By decoupling these processes, FindTrack effectively reduces ambiguities in target association and enhances segmentation consistency. We demonstrate that FindTrack outperforms existing methods on public benchmarks.
☆ Feature Point Extraction for Extra-Affine Image
The issue concerning the significant decline in the stability of feature extraction for images subjected to large-angle affine transformations, where the angle exceeds 50 degrees, still awaits a satisfactory solution. Even ASIFT, which is built upon SIFT and entails a considerable number of image comparisons simulated by affine transformations, inevitably exhibits the drawbacks of being time-consuming and imposing high demands on memory usage. And the stability of feature extraction drops rapidly under large-view affine transformations. Consequently, we propose a method that represents an improvement over ASIFT. On the premise of improving the precision and maintaining the affine invariance, it currently ranks as the fastest feature extraction method for extra-affine images that we know of at present. Simultaneously, the stability of feature extraction regarding affine transformation images has been approximated to the maximum limits. Both the angle between the shooting direction and the normal direction of the photographed object (absolute tilt angle), and the shooting transformation angle between two images (transition tilt angle) are close to 90 degrees. The central idea of the method lies in obtaining the optimal parameter set by simulating affine transformation with the reference image. And the simulated affine transformation is reproduced by combining it with the Lanczos interpolation based on the optimal parameter set. Subsequently, it is combined with ORB, which exhibits excellent real-time performance for rapid orientation binary description. Moreover, a scale parameter simulation is introduced to further augment the operational efficiency.
☆ Bridging Synthetic-to-Real Gaps: Frequency-Aware Perturbation and Selection for Single-shot Multi-Parametric Mapping Reconstruction
Data-centric artificial intelligence (AI) has remarkably advanced medical imaging, with emerging methods using synthetic data to address data scarcity while introducing synthetic-to-real gaps. Unsupervised domain adaptation (UDA) shows promise in ground truth-scarce tasks, but its application in reconstruction remains underexplored. Although multiple overlapping-echo detachment (MOLED) achieves ultra-fast multi-parametric reconstruction, extending its application to various clinical scenarios, the quality suffers from deficiency in mitigating the domain gap, difficulty in maintaining structural integrity, and inadequacy in ensuring mapping accuracy. To resolve these issues, we proposed frequency-aware perturbation and selection (FPS), comprising Wasserstein distance-modulated frequency-aware perturbation (WDFP) and hierarchical frequency-aware selection network (HFSNet), which integrates frequency-aware adaptive selection (FAS), compact FAS (cFAS) and feature-aware architecture integration (FAI). Specifically, perturbation activates domain-invariant feature learning within uncertainty, while selection refines optimal solutions within perturbation, establishing a robust and closed-loop learning pathway. Extensive experiments on synthetic data, along with diverse real clinical cases from 5 healthy volunteers, 94 ischemic stroke patients, and 46 meningioma patients, demonstrate the superiority and clinical applicability of FPS. Furthermore, FPS is applied to diffusion tensor imaging (DTI), underscoring its versatility and potential for broader medical applications. The code is available at https://github.com/flyannie/FPS.
comment: This work will be submitted to the IEEE for possible publication
☆ DTU-Net: A Multi-Scale Dilated Transformer Network for Nonlinear Hyperspectral Unmixing
Transformers have shown significant success in hyperspectral unmixing (HU). However, challenges remain. While multi-scale and long-range spatial correlations are essential in unmixing tasks, current Transformer-based unmixing networks, built on Vision Transformer (ViT) or Swin-Transformer, struggle to capture them effectively. Additionally, current Transformer-based unmixing networks rely on the linear mixing model, which lacks the flexibility to accommodate scenarios where nonlinear effects are significant. To address these limitations, we propose a multi-scale Dilated Transformer-based unmixing network for nonlinear HU (DTU-Net). The encoder employs two branches. The first one performs multi-scale spatial feature extraction using Multi-Scale Dilated Attention (MSDA) in the Dilated Transformer, which varies dilation rates across attention heads to capture long-range and multi-scale spatial correlations. The second one performs spectral feature extraction utilizing 3D-CNNs with channel attention. The outputs from both branches are then fused to integrate multi-scale spatial and spectral information, which is subsequently transformed to estimate the abundances. The decoder is designed to accommodate both linear and nonlinear mixing scenarios. Its interpretability is enhanced by explicitly modeling the relationships between endmembers, abundances, and nonlinear coefficients in accordance with the polynomial post-nonlinear mixing model (PPNMM). Experiments on synthetic and real datasets validate the effectiveness of the proposed DTU-Net compared to PPNMM-derived methods and several advanced unmixing networks.
☆ Active Learning for Deep Learning-Based Hemodynamic Parameter Estimation
Hemodynamic parameters such as pressure and wall shear stress play an important role in diagnosis, prognosis, and treatment planning in cardiovascular diseases. These parameters can be accurately computed using computational fluid dynamics (CFD), but CFD is computationally intensive. Hence, deep learning methods have been adopted as a surrogate to rapidly estimate CFD outcomes. A drawback of such data-driven models is the need for time-consuming reference CFD simulations for training. In this work, we introduce an active learning framework to reduce the number of CFD simulations required for the training of surrogate models, lowering the barriers to their deployment in new applications. We propose three distinct querying strategies to determine for which unlabeled samples CFD simulations should be obtained. These querying strategies are based on geometrical variance, ensemble uncertainty, and adherence to the physics governing fluid dynamics. We benchmark these methods on velocity field estimation in synthetic coronary artery bifurcations and find that they allow for substantial reductions in annotation cost. Notably, we find that our strategies reduce the number of samples required by up to 50% and make the trained models more robust to difficult cases. Our results show that active learning is a feasible strategy to increase the potential of deep learning-based CFD surrogates.
☆ Biased Heritage: How Datasets Shape Models in Facial Expression Recognition
In recent years, the rapid development of artificial intelligence (AI) systems has raised concerns about our ability to ensure their fairness, that is, how to avoid discrimination based on protected characteristics such as gender, race, or age. While algorithmic fairness is well-studied in simple binary classification tasks on tabular data, its application to complex, real-world scenarios-such as Facial Expression Recognition (FER)-remains underexplored. FER presents unique challenges: it is inherently multiclass, and biases emerge across intersecting demographic variables, each potentially comprising multiple protected groups. We present a comprehensive framework to analyze bias propagation from datasets to trained models in image-based FER systems, while introducing new bias metrics specifically designed for multiclass problems with multiple demographic groups. Our methodology studies bias propagation by (1) inducing controlled biases in FER datasets, (2) training models on these biased datasets, and (3) analyzing the correlation between dataset bias metrics and model fairness notions. Our findings reveal that stereotypical biases propagate more strongly to model predictions than representational biases, suggesting that preventing emotion-specific demographic patterns should be prioritized over general demographic balance in FER datasets. Additionally, we observe that biased datasets lead to reduced model accuracy, challenging the assumed fairness-accuracy trade-off.
comment: 17 pages, 7 figures
☆ JamMa: Ultra-lightweight Local Feature Matching with Joint Mamba CVPR 2025
Existing state-of-the-art feature matchers capture long-range dependencies with Transformers but are hindered by high spatial complexity, leading to demanding training and highlatency inference. Striking a better balance between performance and efficiency remains a challenge in feature matching. Inspired by the linear complexity O(N) of Mamba, we propose an ultra-lightweight Mamba-based matcher, named JamMa, which converges on a single GPU and achieves an impressive performance-efficiency balance in inference. To unlock the potential of Mamba for feature matching, we propose Joint Mamba with a scan-merge strategy named JEGO, which enables: (1) Joint scan of two images to achieve high-frequency mutual interaction, (2) Efficient scan with skip steps to reduce sequence length, (3) Global receptive field, and (4) Omnidirectional feature representation. With the above properties, the JEGO strategy significantly outperforms the scan-merge strategies proposed in VMamba and EVMamba in the feature matching task. Compared to attention-based sparse and semi-dense matchers, JamMa demonstrates a superior balance between performance and efficiency, delivering better performance with less than 50% of the parameters and FLOPs.
comment: CVPR 2025, Project page: https://leoluxxx.github.io/JamMa-page/
☆ CoSDH: Communication-Efficient Collaborative Perception via Supply-Demand Awareness and Intermediate-Late Hybridization CVPR 2025
Multi-agent collaborative perception enhances perceptual capabilities by utilizing information from multiple agents and is considered a fundamental solution to the problem of weak single-vehicle perception in autonomous driving. However, existing collaborative perception methods face a dilemma between communication efficiency and perception accuracy. To address this issue, we propose a novel communication-efficient collaborative perception framework based on supply-demand awareness and intermediate-late hybridization, dubbed as \mymethodname. By modeling the supply-demand relationship between agents, the framework refines the selection of collaboration regions, reducing unnecessary communication cost while maintaining accuracy. In addition, we innovatively introduce the intermediate-late hybrid collaboration mode, where late-stage collaboration compensates for the performance degradation in collaborative perception under low communication bandwidth. Extensive experiments on multiple datasets, including both simulated and real-world scenarios, demonstrate that \mymethodname~ achieves state-of-the-art detection accuracy and optimal bandwidth trade-offs, delivering superior detection precision under real communication bandwidths, thus proving its effectiveness and practical applicability. The code will be released at https://github.com/Xu2729/CoSDH.
comment: Accepted at CVPR 2025
☆ Automatic Drywall Analysis for Progress Tracking and Quality Control in Construction
Digitalization in the construction industry has become essential, enabling centralized, easy access to all relevant information of a building. Automated systems can facilitate the timely and resource-efficient documentation of changes, which is crucial for key processes such as progress tracking and quality control. This paper presents a method for image-based automated drywall analysis enabling construction progress and quality assessment through on-site camera systems. Our proposed solution integrates a deep learning-based instance segmentation model to detect and classify various drywall elements with an analysis module to cluster individual wall segments, estimate camera perspective distortions, and apply the corresponding corrections. This system extracts valuable information from images, enabling more accurate progress tracking and quality assessment on construction sites. Our main contributions include a fully automated pipeline for drywall analysis, improving instance segmentation accuracy through architecture modifications and targeted data augmentation, and a novel algorithm to extract important information from the segmentation results. Our modified model, enhanced with data augmentation, achieves significantly higher accuracy compared to other architectures, offering more detailed and precise information than existing approaches. Combined with the proposed drywall analysis steps, it enables the reliable automation of construction progress and quality assessment.
☆ Augmentation-Based Deep Learning for Identification of Circulating Tumor Cells
Circulating tumor cells (CTCs) are crucial biomarkers in liquid biopsy, offering a noninvasive tool for cancer patient management. However, their identification remains particularly challenging due to their limited number and heterogeneity. Labeling samples for contrast limits the generalization of fluorescence-based methods across different hospital datasets. Analyzing single-cell images enables detailed assessment of cell morphology, subcellular structures, and phenotypic variations, often hidden in clustered images. Developing a method based on bright-field single-cell analysis could overcome these limitations. CTCs can be isolated using an unbiased workflow combining Parsortix technology, which selects cells based on size and deformability, with DEPArray technology, enabling precise visualization and selection of single cells. Traditionally, DEPArray-acquired digital images are manually analyzed, making the process time-consuming and prone to variability. In this study, we present a Deep Learning-based classification pipeline designed to distinguish CTCs from leukocytes in blood samples, aimed to enhance diagnostic accuracy and optimize clinical workflows. Our approach employs images from the bright-field channel acquired through DEPArray technology leveraging a ResNet-based CNN. To improve model generalization, we applied three types of data augmentation techniques and incorporated fluorescence (DAPI) channel images into the training phase, allowing the network to learn additional CTC-specific features. Notably, only bright-field images have been used for testing, ensuring the model's ability to identify CTCs without relying on fluorescence markers. The proposed model achieved an F1-score of 0.798, demonstrating its capability to distinguish CTCs from leukocytes. These findings highlight the potential of DL in refining CTC analysis and advancing liquid biopsy applications.
comment: 20 pages, 4 figures, 3 tables
☆ AI-Driven Multi-Stage Computer Vision System for Defect Detection in Laser-Engraved Industrial Nameplates
Automated defect detection in industrial manufacturing is essential for maintaining product quality and minimizing production errors. In air disc brake manufacturing, ensuring the precision of laser-engraved nameplates is crucial for accurate product identification and quality control. Engraving errors, such as misprints or missing characters, can compromise both aesthetics and functionality, leading to material waste and production delays. This paper presents a proof of concept for an AI-driven computer vision system that inspects and verifies laser-engraved nameplates, detecting defects in logos and alphanumeric strings. The system integrates object detection using YOLOv7, optical character recognition (OCR) with Tesseract, and anomaly detection through a residual variational autoencoder (ResVAE) along with other computer vision methods to enable comprehensive inspections at multiple stages. Experimental results demonstrate the system's effectiveness, achieving 91.33% accuracy and 100% recall, ensuring that defective nameplates are consistently detected and addressed. This solution highlights the potential of AI-driven visual inspection to enhance quality control, reduce manual inspection efforts, and improve overall manufacturing efficiency.
☆ MIAdapt: Source-free Few-shot Domain Adaptive Object Detection for Microscopic Images
Existing generic unsupervised domain adaptation approaches require access to both a large labeled source dataset and a sufficient unlabeled target dataset during adaptation. However, collecting a large dataset, even if unlabeled, is a challenging and expensive endeavor, especially in medical imaging. In addition, constraints such as privacy issues can result in cases where source data is unavailable. Taking in consideration these challenges, we propose MIAdapt, an adaptive approach for Microscopic Imagery Adaptation as a solution for Source-free Few-shot Domain Adaptive Object detection (SF-FSDA). We also define two competitive baselines (1) Faster-FreeShot and (2) MT-FreeShot. Extensive experiments on the challenging M5-Malaria and Raabin-WBC datasets validate the effectiveness of MIAdapt. Without using any image from the source domain MIAdapt surpasses state-of-the-art source-free UDA (SF-UDA) methods by +21.3% mAP and few-shot domain adaptation (FSDA) approaches by +4.7% mAP on Raabin-WBC. Our code and models will be publicly available.
comment: Under Review
☆ Top-K Maximum Intensity Projection Priors for 3D Liver Vessel Segmentation
Liver-vessel segmentation is an essential task in the pre-operative planning of liver resection. State-of-the-art 2D or 3D convolution-based methods focusing on liver vessel segmentation on 2D CT cross-sectional views, which do not take into account the global liver-vessel topology. To maintain this global vessel topology, we rely on the underlying physics used in the CT reconstruction process, and apply this to liver-vessel segmentation. Concretely, we introduce the concept of top-k maximum intensity projections, which mimics the CT reconstruction by replacing the integral along each projection direction, with keeping the top-k maxima along each projection direction. We use these top-k maximum projections to condition a diffusion model and generate 3D liver-vessel trees. We evaluate our 3D liver-vessel segmentation on the 3D-ircadb-01 dataset, and achieve the highest Dice coefficient, intersection-over-union (IoU), and Sensitivity scores compared to prior work.
comment: Accepted in 2025 IEEE International Symposium on Biomedical Imaging (ISBI 2025)
☆ TopoMortar: A dataset to evaluate image segmentation methods focused on topology accuracy
We present TopoMortar, a brick wall dataset that is the first dataset specifically designed to evaluate topology-focused image segmentation methods, such as topology loss functions. TopoMortar enables to investigate in two ways whether methods incorporate prior topological knowledge. First, by eliminating challenges seen in real-world data, such as small training set, noisy labels, and out-of-distribution test-set images, that, as we show, impact the effectiveness of topology losses. Second, by allowing to assess in the same dataset topology accuracy across dataset challenges, isolating dataset-related effects from the effect of incorporating prior topological knowledge. In these two experiments, it is deliberately difficult to improve topology accuracy without actually using topology information, thus, permitting to attribute an improvement in topology accuracy to the incorporation of prior topological knowledge. To this end, TopoMortar includes three types of labels (accurate, noisy, pseudo-labels), two fixed training sets (large and small), and in-distribution and out-of-distribution test-set images. We compared eight loss functions on TopoMortar, and we found that clDice achieved the most topologically accurate segmentations, Skeleton Recall loss performed best particularly with noisy labels, and the relative advantageousness of the other loss functions depended on the experimental setting. Additionally, we show that simple methods, such as data augmentation and self-distillation, can elevate Cross entropy Dice loss to surpass most topology loss functions, and that those simple methods can enhance topology loss functions as well. clDice and Skeleton Recall loss, both skeletonization-based loss functions, were also the fastest to train, making this type of loss function a promising research direction. TopoMortar and our code can be found at https://github.com/jmlipman/TopoMortar
☆ Video Super-Resolution: All You Need is a Video Diffusion Model
We present a generic video super-resolution algorithm in this paper, based on the Diffusion Posterior Sampling framework with an unconditional video generation model in latent space. The video generation model, a diffusion transformer, functions as a space-time model. We argue that a powerful model, which learns the physics of the real world, can easily handle various kinds of motion patterns as prior knowledge, thus eliminating the need for explicit estimation of optical flows or motion parameters for pixel alignment. Furthermore, a single instance of the proposed video diffusion transformer model can adapt to different sampling conditions without re-training. Due to limited computational resources and training data, our experiments provide empirical evidence of the algorithm's strong super-resolution capabilities using synthetic data.
☆ Automated Attendee Recognition System for Large-Scale Social Events or Conference Gathering
Manual attendance tracking at large-scale events, such as marriage functions or conferences, is often inefficient and prone to human error. To address this challenge, we propose an automated, cloud-based attendance tracking system that uses cameras mounted at the entrance and exit gates. The mounted cameras continuously capture video and send the video data to cloud services to perform real-time face detection and recognition. Unlike existing solutions, our system accurately identifies attendees even when they are not looking directly at the camera, allowing natural movements, such as looking around or talking while walking. To the best of our knowledge, this is the first system to achieve high recognition rates under such dynamic conditions. Our system demonstrates overall 90% accuracy, with each video frame processed in 5 seconds, ensuring real time operation without frame loss. In addition, notifications are sent promptly to security personnel within the same latency. This system achieves 100% accuracy for individuals without facial obstructions and successfully recognizes all attendees appearing within the camera's field of view, providing a robust solution for attendee recognition in large-scale social events.
☆ Deep Learning-Based Diffusion MRI Tractography: Integrating Spatial and Anatomical Information
Diffusion MRI tractography technique enables non-invasive visualization of the white matter pathways in the brain. It plays a crucial role in neuroscience and clinical fields by facilitating the study of brain connectivity and neurological disorders. However, the accuracy of reconstructed tractograms has been a longstanding challenge. Recently, deep learning methods have been applied to improve tractograms for better white matter coverage, but often comes at the expense of generating excessive false-positive connections. This is largely due to their reliance on local information to predict long range streamlines. To improve the accuracy of streamline propagation predictions, we introduce a novel deep learning framework that integrates image-domain spatial information and anatomical information along tracts, with the former extracted through convolutional layers and the later modeled via a Transformer-decoder. Additionally, we employ a weighted loss function to address fiber class imbalance encountered during training. We evaluate the proposed method on the simulated ISMRM 2015 Tractography Challenge dataset, achieving a valid streamline rate of 66.2%, white matter coverage of 63.8%, and successfully reconstructing 24 out of 25 bundles. Furthermore, on the multi-site Tractoinferno dataset, the proposed method demonstrates its ability to handle various diffusion MRI acquisition schemes, achieving a 5.7% increase in white matter coverage and a 4.1% decrease in overreach compared to RNN-based methods.
☆ ScaleFusionNet: Transformer-Guided Multi-Scale Feature Fusion for Skin Lesion Segmentation
Melanoma is a malignant tumor originating from skin cell lesions. Accurate and efficient segmentation of skin lesions is essential for quantitative medical analysis but remains challenging. To address this, we propose ScaleFusionNet, a segmentation model that integrates Cross-Attention Transformer Module (CATM) and AdaptiveFusionBlock to enhance feature extraction and fusion. The model employs a hybrid architecture encoder that effectively captures both local and global features. We introduce CATM, which utilizes Swin Transformer Blocks and Cross Attention Fusion (CAF) to adaptively refine encoder-decoder feature fusion, reducing semantic gaps and improving segmentation accuracy. Additionally, the AdaptiveFusionBlock is improved by integrating adaptive multi-scale fusion, where Swin Transformer-based attention complements deformable convolution-based multi-scale feature extraction. This enhancement refines lesion boundaries and preserves fine-grained details. ScaleFusionNet achieves Dice scores of 92.94% and 91.65% on ISIC-2016 and ISIC-2018 datasets, respectively, demonstrating its effectiveness in skin lesion analysis. Our code implementation is publicly available at GitHub.
☆ Golden Cudgel Network for Real-Time Semantic Segmentation
Recent real-time semantic segmentation models, whether single-branch or multi-branch, achieve good performance and speed. However, their speed is limited by multi-path blocks, and some depend on high-performance teacher models for training. To overcome these issues, we propose Golden Cudgel Network (GCNet). Specifically, GCNet uses vertical multi-convolutions and horizontal multi-paths for training, which are reparameterized into a single convolution for inference, optimizing both performance and speed. This design allows GCNet to self-enlarge during training and self-contract during inference, effectively becoming a "teacher model" without needing external ones. Experimental results show that GCNet outperforms existing state-of-the-art models in terms of performance and speed on the Cityscapes, CamVid, and Pascal VOC 2012 datasets. The code is available at https://github.com/gyyang23/GCNet.
☆ See What You Are Told: Visual Attention Sink in Large Multimodal Models
Large multimodal models (LMMs) "see" images by leveraging the attention mechanism between text and visual tokens in the transformer decoder. Ideally, these models should focus on key visual information relevant to the text token. However, recent findings indicate that LMMs have an extraordinary tendency to consistently allocate high attention weights to specific visual tokens, even when these tokens are irrelevant to the corresponding text. In this study, we investigate the property behind the appearance of these irrelevant visual tokens and examine their characteristics. Our findings show that this behavior arises due to the massive activation of certain hidden state dimensions, which resembles the attention sink found in language models. Hence, we refer to this phenomenon as the visual attention sink. In particular, our analysis reveals that removing the irrelevant visual sink tokens does not impact model performance, despite receiving high attention weights. Consequently, we recycle the attention to these tokens as surplus resources, redistributing the attention budget to enhance focus on the image. To achieve this, we introduce Visual Attention Redistribution (VAR), a method that redistributes attention in image-centric heads, which we identify as innately focusing on visual information. VAR can be seamlessly applied across different LMMs to improve performance on a wide range of tasks, including general vision-language tasks, visual hallucination tasks, and vision-centric tasks, all without the need for additional training, models, or inference steps. Experimental results demonstrate that VAR enables LMMs to process visual information more effectively by adjusting their internal attention mechanisms, offering a new direction to enhancing the multimodal capabilities of LMMs.
☆ Full-DoF Egomotion Estimation for Event Cameras Using Geometric Solvers CVPR
For event cameras, current sparse geometric solvers for egomotion estimation assume that the rotational displacements are known, such as those provided by an IMU. Thus, they can only recover the translational motion parameters. Recovering full-DoF motion parameters using a sparse geometric solver is a more challenging task, and has not yet been investigated. In this paper, we propose several solvers to estimate both rotational and translational velocities within a unified framework. Our method leverages event manifolds induced by line segments. The problem formulations are based on either an incidence relation for lines or a novel coplanarity relation for normal vectors. We demonstrate the possibility of recovering full-DoF egomotion parameters for both angular and linear velocities without requiring extra sensor measurements or motion priors. To achieve efficient optimization, we exploit the Adam framework with a first-order approximation of rotations for quick initialization. Experiments on both synthetic and real-world data demonstrate the effectiveness of our method. The code is available at https://github.com/jizhaox/relpose-event.
comment: Accepted by IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025
☆ Label-Efficient LiDAR Semantic Segmentation with 2D-3D Vision Transformer Adapters
LiDAR semantic segmentation models are typically trained from random initialization as universal pre-training is hindered by the lack of large, diverse datasets. Moreover, most point cloud segmentation architectures incorporate custom network layers, limiting the transferability of advances from vision-based architectures. Inspired by recent advances in universal foundation models, we propose BALViT, a novel approach that leverages frozen vision models as amodal feature encoders for learning strong LiDAR encoders. Specifically, BALViT incorporates both range-view and bird's-eye-view LiDAR encoding mechanisms, which we combine through a novel 2D-3D adapter. While the range-view features are processed through a frozen image backbone, our bird's-eye-view branch enhances them through multiple cross-attention interactions. Thereby, we continuously improve the vision network with domain-dependent knowledge, resulting in a strong label-efficient LiDAR encoding mechanism. Extensive evaluations of BALViT on the SemanticKITTI and nuScenes benchmarks demonstrate that it outperforms state-of-the-art methods on small data regimes. We make the code and models publicly available at: http://balvit.cs.uni-freiburg.de.
☆ Interactive Segmentation and Report Generation for CT Images
Automated CT report generation plays a crucial role in improving diagnostic accuracy and clinical workflow efficiency. However, existing methods lack interpretability and impede patient-clinician understanding, while their static nature restricts radiologists from dynamically adjusting assessments during image review. Inspired by interactive segmentation techniques, we propose a novel interactive framework for 3D lesion morphology reporting that seamlessly generates segmentation masks with comprehensive attribute descriptions, enabling clinicians to generate detailed lesion profiles for enhanced diagnostic assessment. To our best knowledge, we are the first to integrate the interactive segmentation and structured reports in 3D CT medical images. Experimental results across 15 lesion types demonstrate the effectiveness of our approach in providing a more comprehensive and reliable reporting system for lesion segmentation and capturing. The source code will be made publicly available following paper acceptance.
☆ Deep Understanding of Sign Language for Sign to Subtitle Alignment
The objective of this work is to align asynchronous subtitles in sign language videos with limited labelled data. To achieve this goal, we propose a novel framework with the following contributions: (1) we leverage fundamental grammatical rules of British Sign Language (BSL) to pre-process the input subtitles, (2) we design a selective alignment loss to optimise the model for predicting the temporal location of signs only when the queried sign actually occurs in a scene, and (3) we conduct self-training with refined pseudo-labels which are more accurate than the heuristic audio-aligned labels. From this, our model not only better understands the correlation between the text and the signs, but also holds potential for application in the translation of sign languages, particularly in scenarios where manual labelling of large-scale sign data is impractical or challenging. Extensive experimental results demonstrate that our approach achieves state-of-the-art results, surpassing previous baselines by substantial margins in terms of both frame-level accuracy and F1-score. This highlights the effectiveness and practicality of our framework in advancing the field of sign language video alignment and translation.
☆ Enhancing Visual Forced Alignment with Local Context-Aware Feature Extraction and Multi-Task Learning ICASSP2025
This paper introduces a novel approach to Visual Forced Alignment (VFA), aiming to accurately synchronize utterances with corresponding lip movements, without relying on audio cues. We propose a novel VFA approach that integrates a local context-aware feature extractor and employs multi-task learning to refine both global and local context features, enhancing sensitivity to subtle lip movements for precise word-level and phoneme-level alignment. Incorporating the improved Viterbi algorithm for post-processing, our method significantly reduces misalignments. Experimental results show our approach outperforms existing methods, achieving a 6% accuracy improvement at the word-level and 27% improvement at the phoneme-level in LRS2 dataset. These improvements offer new potential for applications in automatically subtitling TV shows or user-generated content platforms like TikTok and YouTube Shorts.
comment: Accepted by ICASSP2025
☆ Enhancing Vietnamese VQA through Curriculum Learning on Raw and Augmented Text Representations AAAI-25
Visual Question Answering (VQA) is a multimodal task requiring reasoning across textual and visual inputs, which becomes particularly challenging in low-resource languages like Vietnamese due to linguistic variability and the lack of high-quality datasets. Traditional methods often rely heavily on extensive annotated datasets, computationally expensive pipelines, and large pre-trained models, specifically in the domain of Vietnamese VQA, limiting their applicability in such scenarios. To address these limitations, we propose a training framework that combines a paraphrase-based feature augmentation module with a dynamic curriculum learning strategy. Explicitly, augmented samples are considered "easy" while raw samples are regarded as "hard". The framework then utilizes a mechanism that dynamically adjusts the ratio of easy to hard samples during training, progressively modifying the same dataset to increase its difficulty level. By enabling gradual adaptation to task complexity, this approach helps the Vietnamese VQA model generalize well, thus improving overall performance. Experimental results show consistent improvements on the OpenViVQA dataset and mixed outcomes on the ViVQA dataset, highlighting both the potential and challenges of our approach in advancing VQA for Vietnamese language.
comment: 10 pages, 3 figures, AAAI-25 Workshop on Document Understanding and Intelligence
☆ Gaussian highpass guided image filtering
Guided image filtering (GIF) is a popular smoothing technique, in which an additional image is used as a structure guidance for noise removal with edge preservation. The original GIF and some of its subsequent improvements are derived from a two-parameter local affine model (LAM), where the filtering output is a local affine transformation of the guidance image, but the input image is not taken into account in the LAM formulation. In this paper, we first introduce a single-parameter Prior Model based on Gaussian (highpass/lowpass) Filtering (PM-GF), in which the filtering output is the sum of a weighted portion of Gaussian highpass filtering of the guidance image and Gaussian smoothing of the input image. In the PM-GF, the guidance structure determined by Gaussian highpass filtering is obviously transferred to the filtering output, thereby better revealing the structure transfer mechanism of guided filtering. Then we propose several Gaussian highpass GIFs (GH-GIFs) based on the PM-GF by emulating the original GIF and some improvements, i.e., using PM-GF instead of LAM in these GIFs. Experimental results illustrate that the proposed GIFs outperform their counterparts in several image processing applications.
☆ BEVMOSNet: Multimodal Fusion for BEV Moving Object Segmentation
Accurate motion understanding of the dynamic objects within the scene in bird's-eye-view (BEV) is critical to ensure a reliable obstacle avoidance system and smooth path planning for autonomous vehicles. However, this task has received relatively limited exploration when compared to object detection and segmentation with only a few recent vision-based approaches presenting preliminary findings that significantly deteriorate in low-light, nighttime, and adverse weather conditions such as rain. Conversely, LiDAR and radar sensors remain almost unaffected in these scenarios, and radar provides key velocity information of the objects. Therefore, we introduce BEVMOSNet, to our knowledge, the first end-to-end multimodal fusion leveraging cameras, LiDAR, and radar to precisely predict the moving objects in BEV. In addition, we perform a deeper analysis to find out the optimal strategy for deformable cross-attention-guided sensor fusion for cross-sensor knowledge sharing in BEV. While evaluating BEVMOSNet on the nuScenes dataset, we show an overall improvement in IoU score of 36.59% compared to the vision-based unimodal baseline BEV-MoSeg (Sigatapu et al., 2023), and 2.35% compared to the multimodel SimpleBEV (Harley et al., 2022), extended for the motion segmentation task, establishing this method as the state-of-the-art in BEV motion segmentation.
comment: In Proceedings of the 20th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (2025)
☆ Enhancing Abnormality Grounding for Vision Language Models with Knowledge Descriptions
Visual Language Models (VLMs) have demonstrated impressive capabilities in visual grounding tasks. However, their effectiveness in the medical domain, particularly for abnormality detection and localization within medical images, remains underexplored. A major challenge is the complex and abstract nature of medical terminology, which makes it difficult to directly associate pathological anomaly terms with their corresponding visual features. In this work, we introduce a novel approach to enhance VLM performance in medical abnormality detection and localization by leveraging decomposed medical knowledge. Instead of directly prompting models to recognize specific abnormalities, we focus on breaking down medical concepts into fundamental attributes and common visual patterns. This strategy promotes a stronger alignment between textual descriptions and visual features, improving both the recognition and localization of abnormalities in medical images.We evaluate our method on the 0.23B Florence-2 base model and demonstrate that it achieves comparable performance in abnormality grounding to significantly larger 7B LLaVA-based medical VLMs, despite being trained on only 1.5% of the data used for such models. Experimental results also demonstrate the effectiveness of our approach in both known and previously unseen abnormalities, suggesting its strong generalization capabilities.
comment: 11 pages, 3 figures
☆ Towards Effective and Sparse Adversarial Attack on Spiking Neural Networks via Breaking Invisible Surrogate Gradients CVPR 2025
Spiking neural networks (SNNs) have shown their competence in handling spatial-temporal event-based data with low energy consumption. Similar to conventional artificial neural networks (ANNs), SNNs are also vulnerable to gradient-based adversarial attacks, wherein gradients are calculated by spatial-temporal back-propagation (STBP) and surrogate gradients (SGs). However, the SGs may be invisible for an inference-only model as they do not influence the inference results, and current gradient-based attacks are ineffective for binary dynamic images captured by the dynamic vision sensor (DVS). While some approaches addressed the issue of invisible SGs through universal SGs, their SGs lack a correlation with the victim model, resulting in sub-optimal performance. Moreover, the imperceptibility of existing SNN-based binary attacks is still insufficient. In this paper, we introduce an innovative potential-dependent surrogate gradient (PDSG) method to establish a robust connection between the SG and the model, thereby enhancing the adaptability of adversarial attacks across various models with invisible SGs. Additionally, we propose the sparse dynamic attack (SDA) to effectively attack binary dynamic images. Utilizing a generation-reduction paradigm, SDA can fully optimize the sparsity of adversarial perturbations. Experimental results demonstrate that our PDSG and SDA outperform state-of-the-art SNN-based attacks across various models and datasets. Specifically, our PDSG achieves 100% attack success rate on ImageNet, and our SDA obtains 82% attack success rate by modifying only 0.24% of the pixels on CIFAR10DVS. The code is available at https://github.com/ryime/PDSG-SDA .
comment: Accepted by CVPR 2025
☆ Reduced Spatial Dependency for More General Video-level Deepfake Detection ICASSP 2025
As one of the prominent AI-generated content, Deepfake has raised significant safety concerns. Although it has been demonstrated that temporal consistency cues offer better generalization capability, existing methods based on CNNs inevitably introduce spatial bias, which hinders the extraction of intrinsic temporal features. To address this issue, we propose a novel method called Spatial Dependency Reduction (SDR), which integrates common temporal consistency features from multiple spatially-perturbed clusters, to reduce the dependency of the model on spatial information. Specifically, we design multiple Spatial Perturbation Branch (SPB) to construct spatially-perturbed feature clusters. Subsequently, we utilize the theory of mutual information and propose a Task-Relevant Feature Integration (TRFI) module to capture temporal features residing in similar latent space from these clusters. Finally, the integrated feature is fed into a temporal transformer to capture long-range dependencies. Extensive benchmarks and ablation studies demonstrate the effectiveness and rationale of our approach.
comment: 5 pages, 2 figures. Accepted to ICASSP 2025
☆ Optimizing for the Shortest Path in Denoising Diffusion Model CVPR 2025
In this research, we propose a novel denoising diffusion model based on shortest-path modeling that optimizes residual propagation to enhance both denoising efficiency and quality.Drawing on Denoising Diffusion Implicit Models (DDIM) and insights from graph theory, our model, termed the Shortest Path Diffusion Model (ShortDF), treats the denoising process as a shortest-path problem aimed at minimizing reconstruction error. By optimizing the initial residuals, we improve the efficiency of the reverse diffusion process and the quality of the generated samples.Extensive experiments on multiple standard benchmarks demonstrate that ShortDF significantly reduces diffusion time (or steps) while enhancing the visual fidelity of generated samples compared to prior arts.This work, we suppose, paves the way for interactive diffusion-based applications and establishes a foundation for rapid data generation. Code is available at https://github.com/UnicomAI/ShortDF.
comment: Accepet by CVPR 2025 (10 pages, 6 figures)
☆ Trajectory Prediction for Autonomous Driving: Progress, Limitations, and Future Directions
As the potential for autonomous vehicles to be integrated on a large scale into modern traffic systems continues to grow, ensuring safe navigation in dynamic environments is crucial for smooth integration. To guarantee safety and prevent collisions, autonomous vehicles must be capable of accurately predicting the trajectories of surrounding traffic agents. Over the past decade, significant efforts from both academia and industry have been dedicated to designing solutions for precise trajectory forecasting. These efforts have produced a diverse range of approaches, raising questions about the differences between these methods and whether trajectory prediction challenges have been fully addressed. This paper reviews a substantial portion of recent trajectory prediction methods and devises a taxonomy to classify existing solutions. A general overview of the prediction pipeline is also provided, covering input and output modalities, modeling features, and prediction paradigms discussed in the literature. In addition, the paper discusses active research areas within trajectory prediction, addresses the posed research questions, and highlights the remaining research gaps and challenges.
☆ BANet: Bilateral Aggregation Network for Mobile Stereo Matching
State-of-the-art stereo matching methods typically use costly 3D convolutions to aggregate a full cost volume, but their computational demands make mobile deployment challenging. Directly applying 2D convolutions for cost aggregation often results in edge blurring, detail loss, and mismatches in textureless regions. Some complex operations, like deformable convolutions and iterative warping, can partially alleviate this issue; however, they are not mobile-friendly, limiting their deployment on mobile devices. In this paper, we present a novel bilateral aggregation network (BANet) for mobile stereo matching that produces high-quality results with sharp edges and fine details using only 2D convolutions. Specifically, we first separate the full cost volume into detailed and smooth volumes using a spatial attention map, then perform detailed and smooth aggregations accordingly, ultimately fusing both to obtain the final disparity map. Additionally, to accurately identify high-frequency detailed regions and low-frequency smooth/textureless regions, we propose a new scale-aware spatial attention module. Experimental results demonstrate that our BANet-2D significantly outperforms other mobile-friendly methods, achieving 35.3\% higher accuracy on the KITTI 2015 leaderboard than MobileStereoNet-2D, with faster runtime on mobile devices. The extended 3D version, BANet-3D, achieves the highest accuracy among all real-time methods on high-end GPUs. Code: \textcolor{magenta}{https://github.com/gangweiX/BANet}.
comment: 12 pages
☆ BAT: Learning Event-based Optical Flow with Bidirectional Adaptive Temporal Correlation
Event cameras deliver visual information characterized by a high dynamic range and high temporal resolution, offering significant advantages in estimating optical flow for complex lighting conditions and fast-moving objects. Current advanced optical flow methods for event cameras largely adopt established image-based frameworks. However, the spatial sparsity of event data limits their performance. In this paper, we present BAT, an innovative framework that estimates event-based optical flow using bidirectional adaptive temporal correlation. BAT includes three novel designs: 1) a bidirectional temporal correlation that transforms bidirectional temporally dense motion cues into spatially dense ones, enabling accurate and spatially dense optical flow estimation; 2) an adaptive temporal sampling strategy for maintaining temporal consistency in correlation; 3) spatially adaptive temporal motion aggregation to efficiently and adaptively aggregate consistent target motion features into adjacent motion features while suppressing inconsistent ones. Our results rank $1^{st}$ on the DSEC-Flow benchmark, outperforming existing state-of-the-art methods by a large margin while also exhibiting sharp edges and high-quality details. Notably, our BAT can accurately predict future optical flow using only past events, significantly outperforming E-RAFT's warm-start approach. Code: \textcolor{magenta}{https://github.com/gangweiX/BAT}.
comment: 10 pages
☆ Computational Analysis of Degradation Modeling in Blind Panoramic Image Quality Assessment
Blind panoramic image quality assessment (BPIQA) has recently brought new challenge to the visual quality community, due to the complex interaction between immersive content and human behavior. Although many efforts have been made to advance BPIQA from both conducting psychophysical experiments and designing performance-driven objective algorithms, \textit{limited content} and \textit{few samples} in those closed sets inevitably would result in shaky conclusions, thereby hindering the development of BPIQA, we refer to it as the \textit{easy-database} issue. In this paper, we present a sufficient computational analysis of degradation modeling in BPIQA to thoroughly explore the \textit{easy-database issue}, where we carefully design three types of experiments via investigating the gap between BPIQA and blind image quality assessment (BIQA), the necessity of specific design in BPIQA models, and the generalization ability of BPIQA models. From extensive experiments, we find that easy databases narrow the gap between the performance of BPIQA and BIQA models, which is unconducive to the development of BPIQA. And the easy databases make the BPIQA models be closed to saturation, therefore the effectiveness of the associated specific designs can not be well verified. Besides, the BPIQA models trained on our recently proposed databases with complicated degradation show better generalization ability. Thus, we believe that much more efforts are highly desired to put into BPIQA from both subjective viewpoint and objective viewpoint.
☆ Two-Stream Thermal Imaging Fusion for Enhanced Time of Birth Detection in Neonatal Care
Around 10% of newborns require some help to initiate breathing, and 5\% need ventilation assistance. Accurate Time of Birth (ToB) documentation is essential for optimizing neonatal care, as timely interventions are vital for proper resuscitation. However, current clinical methods for recording ToB often rely on manual processes, which can be prone to inaccuracies. In this study, we present a novel two-stream fusion system that combines the power of image and video analysis to accurately detect the ToB from thermal recordings in the delivery room and operating theater. By integrating static and dynamic streams, our approach captures richer birth-related spatiotemporal features, leading to more robust and precise ToB estimation. We demonstrate that this synergy between data modalities enhances performance over single-stream approaches. Our system achieves 95.7% precision and 84.8% recall in detecting birth within short video clips. Additionally, with the help of a score aggregation module, it successfully identifies ToB in 100% of test cases, with a median absolute error of 2 seconds and an absolute mean deviation of 4.5 seconds compared to manual annotations.
comment: Submitted to IEEE 25th International Conference on Digital Signal Processing
☆ GenColor: Generative Color-Concept Association in Visual Design
Existing approaches for color-concept association typically rely on query-based image referencing, and color extraction from image references. However, these approaches are effective only for common concepts, and are vulnerable to unstable image referencing and varying image conditions. Our formative study with designers underscores the need for primary-accent color compositions and context-dependent colors (e.g., 'clear' vs. 'polluted' sky) in design. In response, we introduce a generative approach for mining semantically resonant colors leveraging images generated by text-to-image models. Our insight is that contemporary text-to-image models can resemble visual patterns from large-scale real-world data. The framework comprises three stages: concept instancing produces generative samples using diffusion models, text-guided image segmentation identifies concept-relevant regions within the image, and color association extracts primarily accompanied by accent colors. Quantitative comparisons with expert designs validate our approach's effectiveness, and we demonstrate the applicability through cases in various design scenarios and a gallery.
comment: 19 pages, 16 figures. Accepted at CHI Conference on Human Factors in Computing Systems (CHI'25), April 26-May 1, 2025, Yokohama, Japan
☆ Path-Adaptive Matting for Efficient Inference Under Various Computational Cost Constraints AAAI 2025
In this paper, we explore a novel image matting task aimed at achieving efficient inference under various computational cost constraints, specifically FLOP limitations, using a single matting network. Existing matting methods which have not explored scalable architectures or path-learning strategies, fail to tackle this challenge. To overcome these limitations, we introduce Path-Adaptive Matting (PAM), a framework that dynamically adjusts network paths based on image contexts and computational cost constraints. We formulate the training of the computational cost-constrained matting network as a bilevel optimization problem, jointly optimizing the matting network and the path estimator. Building on this formalization, we design a path-adaptive matting architecture by incorporating path selection layers and learnable connect layers to estimate optimal paths and perform efficient inference within a unified network. Furthermore, we propose a performance-aware path-learning strategy to generate path labels online by evaluating a few paths sampled from the prior distribution of optimal paths and network estimations, enabling robust and efficient online path learning. Experiments on five image matting datasets demonstrate that the proposed PAM framework achieves competitive performance across a range of computational cost constraints.
comment: Accepted to AAAI 2025
☆ Mocap-2-to-3: Lifting 2D Diffusion-Based Pretrained Models for 3D Motion Capture
Recovering absolute poses in the world coordinate system from monocular views presents significant challenges. Two primary issues arise in this context. Firstly, existing methods rely on 3D motion data for training, which requires collection in limited environments. Acquiring such 3D labels for new actions in a timely manner is impractical, severely restricting the model's generalization capabilities. In contrast, 2D poses are far more accessible and easier to obtain. Secondly, estimating a person's absolute position in metric space from a single viewpoint is inherently more complex. To address these challenges, we introduce Mocap-2-to-3, a novel framework that decomposes intricate 3D motions into 2D poses, leveraging 2D data to enhance 3D motion reconstruction in diverse scenarios and accurately predict absolute positions in the world coordinate system. We initially pretrain a single-view diffusion model with extensive 2D data, followed by fine-tuning a multi-view diffusion model for view consistency using publicly available 3D data. This strategy facilitates the effective use of large-scale 2D data. Additionally, we propose an innovative human motion representation that decouples local actions from global movements and encodes geometric priors of the ground, ensuring the generative model learns accurate motion priors from 2D data. During inference, this allows for the gradual recovery of global movements, resulting in more plausible positioning. We evaluate our model's performance on real-world datasets, demonstrating superior accuracy in motion and absolute human positioning compared to state-of-the-art methods, along with enhanced generalization and scalability. Our code will be made publicly available.
☆ Rice Grain Size Measurement using Image Processing
The rice grain quality can be determined from its size and chalkiness. The traditional approach to measure the rice grain size involves manual inspection, which is inefficient and leads to inconsistent results. To address this issue, an image processing based approach is proposed and developed in this research. The approach takes image of rice grains as input and outputs the number of rice grains and size of each rice grain. The different steps, such as extraction of region of interest, segmentation of rice grains, and sub-contours removal, involved in the proposed approach are discussed. The approach was tested on rice grain images captured from different height using mobile phone camera. The obtained results show that the proposed approach successfully detected 95\% of the rice grains and achieved 90\% accuracy for length and width measurement.
☆ An Analytical Theory of Power Law Spectral Bias in the Learning Dynamics of Diffusion Models
We developed an analytical framework for understanding how the learned distribution evolves during diffusion model training. Leveraging the Gaussian equivalence principle, we derived exact solutions for the gradient-flow dynamics of weights in one- or two-layer linear denoiser settings with arbitrary data. Remarkably, these solutions allowed us to derive the generated distribution in closed form and its KL divergence through training. These analytical results expose a pronounced power-law spectral bias, i.e., for weights and distributions, the convergence time of a mode follows an inverse power law of its variance. Empirical experiments on both Gaussian and image datasets demonstrate that the power-law spectral bias remains robust even when using deeper or convolutional architectures. Our results underscore the importance of the data covariance in dictating the order and rate at which diffusion models learn different modes of the data, providing potential explanations for why earlier stopping could lead to incorrect details in image generative models.
comment: 50 pages, 10 figures. Preprint
☆ Find Matching Faces Based On Face Parameters
This paper presents an innovative approach that enables the user to find matching faces based on the user-selected face parameters. Through gradio-based user interface, the users can interactively select the face parameters they want in their desired partner. These user-selected face parameters are transformed into a text prompt which is used by the Text-To-Image generation model to generate a realistic face image. Further, the generated image along with the images downloaded from the Jeevansathi.com are processed through face detection and feature extraction model, which results in high dimensional vector embedding of 512 dimensions. The vector embeddings generated from the downloaded images are stored into vector database. Now, the similarity search is carried out between the vector embedding of generated image and the stored vector embeddings. As a result, it displays the top five similar faces based on the user-selected face parameters. This contribution holds a significant potential to turn into a high-quality personalized face matching tool.
☆ Variance-Aware Loss Scheduling for Multimodal Alignment in Low-Data Settings
Training vision-language models for image-text alignment typically requires large datasets to achieve robust performance. In low-data scenarios, standard contrastive learning can struggle to align modalities effectively due to overfitting and unstable training dynamics. In this paper, we propose a variance-aware loss scheduling approach that dynamically adjusts the weighting of the contrastive loss based on the statistical variability (uncertainty) in the model's alignment predictions. Using a subset of the Flickr8k image-caption dataset to simulate limited data conditions, we demonstrate that our approach improves image-text retrieval accuracy compared to a fixed-weight baseline. We also compare against other adaptive weighting strategies (using output entropy and cosine similarity spread) and find that variance-aware scheduling provides the best overall trade-off. Qualitatively, our method yields more distinct multimodal embeddings as shown by t-SNE visualizations. Moreover, in a stress test with noise-injected captions and images, the variance-guided loss proves more robust, maintaining higher recall when random perturbations are introduced. These results highlight the benefit of adaptive loss weighting for multimodal alignment in low-data regimes.
comment: 8 pages, 4 figures
Transformer-Based Spatio-Temporal Association of Apple Fruitlets
In this paper, we present a transformer-based method to spatio-temporally associate apple fruitlets in stereo-images collected on different days and from different camera poses. State-of-the-art association methods in agriculture are dedicated towards matching larger crops using either high-resolution point clouds or temporally stable features, which are both difficult to obtain for smaller fruit in the field. To address these challenges, we propose a transformer-based architecture that encodes the shape and position of each fruitlet, and propagates and refines these features through a series of transformer encoder layers with alternating self and cross-attention. We demonstrate that our method is able to achieve an F1-score of 92.4% on data collected in a commercial apple orchard and outperforms all baselines and ablations.
☆ SpiritSight Agent: Advanced GUI Agent with One Look CVPR 2025
Graphical User Interface (GUI) agents show amazing abilities in assisting human-computer interaction, automating human user's navigation on digital devices. An ideal GUI agent is expected to achieve high accuracy, low latency, and compatibility for different GUI platforms. Recent vision-based approaches have shown promise by leveraging advanced Vision Language Models (VLMs). While they generally meet the requirements of compatibility and low latency, these vision-based GUI agents tend to have low accuracy due to their limitations in element grounding. To address this issue, we propose $\textbf{SpiritSight}$, a vision-based, end-to-end GUI agent that excels in GUI navigation tasks across various GUI platforms. First, we create a multi-level, large-scale, high-quality GUI dataset called $\textbf{GUI-Lasagne}$ using scalable methods, empowering SpiritSight with robust GUI understanding and grounding capabilities. Second, we introduce the $\textbf{Universal Block Parsing (UBP)}$ method to resolve the ambiguity problem in dynamic high-resolution of visual inputs, further enhancing SpiritSight's ability to ground GUI objects. Through these efforts, SpiritSight agent outperforms other advanced methods on diverse GUI benchmarks, demonstrating its superior capability and compatibility in GUI navigation tasks. Models are available at $\href{https://huggingface.co/SenseLLM/SpiritSight-Agent-8B}{this\ URL}$.
comment: Paper accepted to CVPR 2025
☆ DSPNet: Dual-vision Scene Perception for Robust 3D Question Answering
3D Question Answering (3D QA) requires the model to comprehensively understand its situated 3D scene described by the text, then reason about its surrounding environment and answer a question under that situation. However, existing methods usually rely on global scene perception from pure 3D point clouds and overlook the importance of rich local texture details from multi-view images. Moreover, due to the inherent noise in camera poses and complex occlusions, there exists significant feature degradation and reduced feature robustness problems when aligning 3D point cloud with multi-view images. In this paper, we propose a Dual-vision Scene Perception Network (DSPNet), to comprehensively integrate multi-view and point cloud features to improve robustness in 3D QA. Our Text-guided Multi-view Fusion (TGMF) module prioritizes image views that closely match the semantic content of the text. To adaptively fuse back-projected multi-view images with point cloud features, we design the Adaptive Dual-vision Perception (ADVP) module, enhancing 3D scene comprehension. Additionally, our Multimodal Context-guided Reasoning (MCGR) module facilitates robust reasoning by integrating contextual information across visual and linguistic modalities. Experimental results on SQA3D and ScanQA datasets demonstrate the superiority of our DSPNet. Codes will be available at https://github.com/LZ-CH/DSPNet.
☆ Partial Convolution Meets Visual Attention
Designing an efficient and effective neural network has remained a prominent topic in computer vision research. Depthwise onvolution (DWConv) is widely used in efficient CNNs or ViTs, but it needs frequent memory access during inference, which leads to low throughput. FasterNet attempts to introduce partial convolution (PConv) as an alternative to DWConv but compromises the accuracy due to underutilized channels. To remedy this shortcoming and consider the redundancy between feature map channels, we introduce a novel Partial visual ATtention mechanism (PAT) that can efficiently combine PConv with visual attention. Our exploration indicates that the partial attention mechanism can completely replace the full attention mechanism and reduce model parameters and FLOPs. Our PAT can derive three types of blocks: Partial Channel-Attention block (PAT_ch), Partial Spatial-Attention block (PAT_sp) and Partial Self-Attention block (PAT_sf). First, PAT_ch integrates the enhanced Gaussian channel attention mechanism to infuse global distribution information into the untouched channels of PConv. Second, we introduce the spatial-wise attention to the MLP layer to further improve model accuracy. Finally, we replace PAT_ch in the last stage with the self-attention mechanism to extend the global receptive field. Building upon PAT, we propose a novel hybrid network family, named PATNet, which achieves superior top-1 accuracy and inference speed compared to FasterNet on ImageNet-1K classification and excel in both detection and segmentation on the COCO dataset. Particularly, our PATNet-T2 achieves 1.3% higher accuracy than FasterNet-T2, while exhibiting 25% higher GPU throughput and 24% lower CPU latency.
comment: arXiv admin note: substantial text overlap with arXiv:2502.01303
☆ Temporal Separation with Entropy Regularization for Knowledge Distillation in Spiking Neural Networks CVPR 2025
Spiking Neural Networks (SNNs), inspired by the human brain, offer significant computational efficiency through discrete spike-based information transfer. Despite their potential to reduce inference energy consumption, a performance gap persists between SNNs and Artificial Neural Networks (ANNs), primarily due to current training methods and inherent model limitations. While recent research has aimed to enhance SNN learning by employing knowledge distillation (KD) from ANN teacher networks, traditional distillation techniques often overlook the distinctive spatiotemporal properties of SNNs, thus failing to fully leverage their advantages. To overcome these challenge, we propose a novel logit distillation method characterized by temporal separation and entropy regularization. This approach improves existing SNN distillation techniques by performing distillation learning on logits across different time steps, rather than merely on aggregated output features. Furthermore, the integration of entropy regularization stabilizes model optimization and further boosts the performance. Extensive experimental results indicate that our method surpasses prior SNN distillation strategies, whether based on logit distillation, feature distillation, or a combination of both. The code will be available on GitHub.
comment: Accepted by CVPR 2025
☆ Implicit U-KAN2.0: Dynamic, Efficient and Interpretable Medical Image Segmentation
Image segmentation is a fundamental task in both image analysis and medical applications. State-of-the-art methods predominantly rely on encoder-decoder architectures with a U-shaped design, commonly referred to as U-Net. Recent advancements integrating transformers and MLPs improve performance but still face key limitations, such as poor interpretability, difficulty handling intrinsic noise, and constrained expressiveness due to discrete layer structures, often lacking a solid theoretical foundation.In this work, we introduce Implicit U-KAN 2.0, a novel U-Net variant that adopts a two-phase encoder-decoder structure. In the SONO phase, we use a second-order neural ordinary differential equation (NODEs), called the SONO block, for a more efficient, expressive, and theoretically grounded modeling approach. In the SONO-MultiKAN phase, we integrate the second-order NODEs and MultiKAN layer as the core computational block to enhance interpretability and representation power. Our contributions are threefold. First, U-KAN 2.0 is an implicit deep neural network incorporating MultiKAN and second order NODEs, improving interpretability and performance while reducing computational costs. Second, we provide a theoretical analysis demonstrating that the approximation ability of the MultiKAN block is independent of the input dimension. Third, we conduct extensive experiments on a variety of 2D and a single 3D dataset, demonstrating that our model consistently outperforms existing segmentation networks.
☆ Dynamic Neural Surfaces for Elastic 4D Shape Representation and Analysis
We propose a novel framework for the statistical analysis of genus-zero 4D surfaces, i.e., 3D surfaces that deform and evolve over time. This problem is particularly challenging due to the arbitrary parameterizations of these surfaces and their varying deformation speeds, necessitating effective spatiotemporal registration. Traditionally, 4D surfaces are discretized, in space and time, before computing their spatiotemporal registrations, geodesics, and statistics. However, this approach may result in suboptimal solutions and, as we demonstrate in this paper, is not necessary. In contrast, we treat 4D surfaces as continuous functions in both space and time. We introduce Dynamic Spherical Neural Surfaces (D-SNS), an efficient smooth and continuous spatiotemporal representation for genus-0 4D surfaces. We then demonstrate how to perform core 4D shape analysis tasks such as spatiotemporal registration, geodesics computation, and mean 4D shape estimation, directly on these continuous representations without upfront discretization and meshing. By integrating neural representations with classical Riemannian geometry and statistical shape analysis techniques, we provide the building blocks for enabling full functional shape analysis. We demonstrate the efficiency of the framework on 4D human and face datasets. The source code and additional results are available at https://4d-dsns.github.io/DSNS/.
comment: 22 pages, 23 figures, conference paper
☆ NTR-Gaussian: Nighttime Dynamic Thermal Reconstruction with 4D Gaussian Splatting Based on Thermodynamics
Thermal infrared imaging offers the advantage of all-weather capability, enabling non-intrusive measurement of an object's surface temperature. Consequently, thermal infrared images are employed to reconstruct 3D models that accurately reflect the temperature distribution of a scene, aiding in applications such as building monitoring and energy management. However, existing approaches predominantly focus on static 3D reconstruction for a single time period, overlooking the impact of environmental factors on thermal radiation and failing to predict or analyze temperature variations over time. To address these challenges, we propose the NTR-Gaussian method, which treats temperature as a form of thermal radiation, incorporating elements like convective heat transfer and radiative heat dissipation. Our approach utilizes neural networks to predict thermodynamic parameters such as emissivity, convective heat transfer coefficient, and heat capacity. By integrating these predictions, we can accurately forecast thermal temperatures at various times throughout a nighttime scene. Furthermore, we introduce a dynamic dataset specifically for nighttime thermal imagery. Extensive experiments and evaluations demonstrate that NTR-Gaussian significantly outperforms comparison methods in thermal reconstruction, achieving a predicted temperature error within 1 degree Celsius.
comment: IEEE Conference on Computer Vision and Pattern Recognition 2025
☆ An Improved Pure Fully Connected Neural Network for Rice Grain Classification
Rice is a staple food for a significant portion of the world's population, providing essential nutrients and serving as a versatile in-gredient in a wide range of culinary traditions. Recently, the use of deep learning has enabled automated classification of rice, im-proving accuracy and efficiency. However, classical models based on first-stage training may face difficulties in distinguishing between rice varieties with similar external characteristics, thus leading to misclassifications. Considering the transparency and feasibility of model, we selected and gradually improved pure fully connected neural network to achieve classification of rice grain. The dataset we used contains both global and domestic rice images obtained from websites and laboratories respectively. First, the training mode was changed from one-stage training to two-stage training, which significantly contributes to distinguishing two similar types of rice. Secondly, the preprocessing method was changed from random tilting to horizontal or vertical position cor-rection. After those two enhancements, the accuracy of our model increased notably from 97% to 99%. In summary, two subtle methods proposed in this study can remarkably enhance the classification ability of deep learning models in terms of the classification of rice grain.
☆ WarmFed: Federated Learning with Warm-Start for Globalization and Personalization Via Personalized Diffusion Models
Federated Learning (FL) stands as a prominent distributed learning paradigm among multiple clients to achieve a unified global model without privacy leakage. In contrast to FL, Personalized federated learning aims at serving for each client in achieving persoanlized model. However, previous FL frameworks have grappled with a dilemma: the choice between developing a singular global model at the server to bolster globalization or nurturing personalized model at the client to accommodate personalization. Instead of making trade-offs, this paper commences its discourse from the pre-trained initialization, obtaining resilient global information and facilitating the development of both global and personalized models. Specifically, we propose a novel method called WarmFed to achieve this. WarmFed customizes Warm-start through personalized diffusion models, which are generated by local efficient fine-tunining (LoRA). Building upon the Warm-Start, we advance a server-side fine-tuning strategy to derive the global model, and propose a dynamic self-distillation (DSD) to procure more resilient personalized models simultaneously. Comprehensive experiments underscore the substantial gains of our approach across both global and personalized models, achieved within just one-shot and five communication(s).
☆ RVAFM: Re-parameterizing Vertical Attention Fusion Module for Handwritten Paragraph Text Recognition
Handwritten Paragraph Text Recognition (HPTR) is a challenging task in Computer Vision, requiring the transformation of a paragraph text image, rich in handwritten text, into text encoding sequences. One of the most advanced models for this task is Vertical Attention Network (VAN), which utilizes a Vertical Attention Module (VAM) to implicitly segment paragraph text images into text lines, thereby reducing the difficulty of the recognition task. However, from a network structure perspective, VAM is a single-branch module, which is less effective in learning compared to multi-branch modules. In this paper, we propose a new module, named Re-parameterizing Vertical Attention Fusion Module (RVAFM), which incorporates structural re-parameterization techniques. RVAFM decouples the structure of the module during training and inference stages. During training, it uses a multi-branch structure for more effective learning, and during inference, it uses a single-branch structure for faster processing. The features learned by the multi-branch structure are fused into the single-branch structure through a special fusion method named Re-parameterization Fusion (RF) without any loss of information. As a result, we achieve a Character Error Rate (CER) of 4.44% and a Word Error Rate (WER) of 14.37% on the IAM paragraph-level test set. Additionally, the inference speed is slightly faster than VAN.
☆ AHCPTQ: Accurate and Hardware-Compatible Post-Training Quantization for Segment Anything Model
The Segment Anything Model (SAM) has demonstrated strong versatility across various visual tasks. However, its large storage requirements and high computational cost pose challenges for practical deployment. Post-training quantization (PTQ) has emerged as an effective strategy for efficient deployment, but we identify two key challenges in SAM that hinder the effectiveness of existing PTQ methods: the heavy-tailed and skewed distribution of post-GELU activations, and significant inter-channel variation in linear projection activations. To address these challenges, we propose AHCPTQ, an accurate and hardware-efficient PTQ method for SAM. AHCPTQ introduces hardware-compatible Hybrid Log-Uniform Quantization (HLUQ) to manage post-GELU activations, employing log2 quantization for dense small values and uniform quantization for sparse large values to enhance quantization resolution. Additionally, AHCPTQ incorporates Channel-Aware Grouping (CAG) to mitigate inter-channel variation by progressively clustering activation channels with similar distributions, enabling them to share quantization parameters and improving hardware efficiency. The combination of HLUQ and CAG not only enhances quantization effectiveness but also ensures compatibility with efficient hardware execution. For instance, under the W4A4 configuration on the SAM-L model, AHCPTQ achieves 36.6% mAP on instance segmentation with the DINO detector, while achieving a 7.89x speedup and 8.64x energy efficiency over its floating-point counterpart in FPGA implementation.
☆ BEVDriver: Leveraging BEV Maps in LLMs for Robust Closed-Loop Driving
Autonomous driving has the potential to set the stage for more efficient future mobility, requiring the research domain to establish trust through safe, reliable and transparent driving. Large Language Models (LLMs) possess reasoning capabilities and natural language understanding, presenting the potential to serve as generalized decision-makers for ego-motion planning that can interact with humans and navigate environments designed for human drivers. While this research avenue is promising, current autonomous driving approaches are challenged by combining 3D spatial grounding and the reasoning and language capabilities of LLMs. We introduce BEVDriver, an LLM-based model for end-to-end closed-loop driving in CARLA that utilizes latent BEV features as perception input. BEVDriver includes a BEV encoder to efficiently process multi-view images and 3D LiDAR point clouds. Within a common latent space, the BEV features are propagated through a Q-Former to align with natural language instructions and passed to the LLM that predicts and plans precise future trajectories while considering navigation instructions and critical scenarios. On the LangAuto benchmark, our model reaches up to 18.9% higher performance on the Driving Score compared to SoTA methods.
comment: This work has been submitted to the IEEE for possible publication
☆ Multi-View Depth Consistent Image Generation Using Generative AI Models: Application on Architectural Design of University Buildings
In the early stages of architectural design, shoebox models are typically used as a simplified representation of building structures but require extensive operations to transform them into detailed designs. Generative artificial intelligence (AI) provides a promising solution to automate this transformation, but ensuring multi-view consistency remains a significant challenge. To solve this issue, we propose a novel three-stage consistent image generation framework using generative AI models to generate architectural designs from shoebox model representations. The proposed method enhances state-of-the-art image generation diffusion models to generate multi-view consistent architectural images. We employ ControlNet as the backbone and optimize it to accommodate multi-view inputs of architectural shoebox models captured from predefined perspectives. To ensure stylistic and structural consistency across multi-view images, we propose an image space loss module that incorporates style loss, structural loss and angle alignment loss. We then use depth estimation method to extract depth maps from the generated multi-view images. Finally, we use the paired data of the architectural images and depth maps as inputs to improve the multi-view consistency via the depth-aware 3D attention module. Experimental results demonstrate that the proposed framework can generate multi-view architectural images with consistent style and structural coherence from shoebox model inputs.
comment: 10 pages, 7 figures, in Proceedings of CAADRIA2025
♻ ☆ ArcPro: Architectural Programs for Structured 3D Abstraction of Sparse Points CVPR 2025
We introduce ArcPro, a novel learning framework built on architectural programs to recover structured 3D abstractions from highly sparse and low-quality point clouds. Specifically, we design a domain-specific language (DSL) to hierarchically represent building structures as a program, which can be efficiently converted into a mesh. We bridge feedforward and inverse procedural modeling by using a feedforward process for training data synthesis, allowing the network to make reverse predictions. We train an encoder-decoder on the points-program pairs to establish a mapping from unstructured point clouds to architectural programs, where a 3D convolutional encoder extracts point cloud features and a transformer decoder autoregressively predicts the programs in a tokenized form. Inference by our method is highly efficient and produces plausible and faithful 3D abstractions. Comprehensive experiments demonstrate that ArcPro outperforms both traditional architectural proxy reconstruction and learning-based abstraction methods. We further explore its potential to work with multi-view image and natural language inputs.
comment: CVPR 2025 (Patent Protected); Project page: https://vcc.tech/research/2025/ArcPro
♻ ☆ Near-infrared Image Deblurring and Event Denoising with Synergistic Neuromorphic Imaging
The fields of imaging in the nighttime dynamic and other extremely dark conditions have seen impressive and transformative advancements in recent years, partly driven by the rise of novel sensing approaches, e.g., near-infrared (NIR) cameras with high sensitivity and event cameras with minimal blur. However, inappropriate exposure ratios of near-infrared cameras make them susceptible to distortion and blur. Event cameras are also highly sensitive to weak signals at night yet prone to interference, often generating substantial noise and significantly degrading observations and analysis. Herein, we develop a new framework for low-light imaging combined with NIR imaging and event-based techniques, named synergistic neuromorphic imaging, which can jointly achieve NIR image deblurring and event denoising. Harnessing cross-modal features of NIR images and visible events via spectral consistency and higher-order interaction, the NIR images and events are simultaneously fused, enhanced, and bootstrapped. Experiments on real and realistically simulated sequences demonstrate the effectiveness of our method and indicate better accuracy and robustness than other methods in practical scenarios. This study gives impetus to enhance both NIR images and events, which paves the way for high-fidelity low-light imaging and neuromorphic reasoning.
♻ ☆ STAA-SNN: Spatial-Temporal Attention Aggregator for Spiking Neural Networks CVPR 2025
Spiking Neural Networks (SNNs) have gained significant attention due to their biological plausibility and energy efficiency, making them promising alternatives to Artificial Neural Networks (ANNs). However, the performance gap between SNNs and ANNs remains a substantial challenge hindering the widespread adoption of SNNs. In this paper, we propose a Spatial-Temporal Attention Aggregator SNN (STAA-SNN) framework, which dynamically focuses on and captures both spatial and temporal dependencies. First, we introduce a spike-driven self-attention mechanism specifically designed for SNNs. Additionally, we pioneeringly incorporate position encoding to integrate latent temporal relationships into the incoming features. For spatial-temporal information aggregation, we employ step attention to selectively amplify relevant features at different steps. Finally, we implement a time-step random dropout strategy to avoid local optima. As a result, STAA-SNN effectively captures both spatial and temporal dependencies, enabling the model to analyze complex patterns and make accurate predictions. The framework demonstrates exceptional performance across diverse datasets and exhibits strong generalization capabilities. Notably, STAA-SNN achieves state-of-the-art results on neuromorphic datasets CIFAR10-DVS, with remarkable performances of 97.14%, 82.05% and 70.40% on the static datasets CIFAR-10, CIFAR-100 and ImageNet, respectively. Furthermore, our model exhibits improved performance ranging from 0.33\% to 2.80\% with fewer time steps. The code for the model is available on GitHub.
comment: Accepted by CVPR 2025
♻ ☆ CMMLoc: Advancing Text-to-PointCloud Localization with Cauchy-Mixture-Model Based Framework CVPR 2025
The goal of point cloud localization based on linguistic description is to identify a 3D position using textual description in large urban environments, which has potential applications in various fields, such as determining the location for vehicle pickup or goods delivery. Ideally, for a textual description and its corresponding 3D location, the objects around the 3D location should be fully described in the text description. However, in practical scenarios, e.g., vehicle pickup, passengers usually describe only the part of the most significant and nearby surroundings instead of the entire environment. In response to this $\textbf{partially relevant}$ challenge, we propose $\textbf{CMMLoc}$, an uncertainty-aware $\textbf{C}$auchy-$\textbf{M}$ixture-$\textbf{M}$odel ($\textbf{CMM}$) based framework for text-to-point-cloud $\textbf{Loc}$alization. To model the uncertain semantic relations between text and point cloud, we integrate CMM constraints as a prior during the interaction between the two modalities. We further design a spatial consolidation scheme to enable adaptive aggregation of different 3D objects with varying receptive fields. To achieve precise localization, we propose a cardinal direction integration module alongside a modality pre-alignment strategy, helping capture the spatial relationships among objects and bringing the 3D objects closer to the text modality. Comprehensive experiments validate that CMMLoc outperforms existing methods, achieving state-of-the-art results on the KITTI360Pose dataset. Codes are available in this GitHub repository https://github.com/kevin301342/CMMLoc.
comment: Accepted by CVPR 2025
♻ ☆ A Multi-Sensor Fusion Approach for Rapid Orthoimage Generation in Large-Scale UAV Mapping
Rapid generation of large-scale orthoimages from Unmanned Aerial Vehicles (UAVs) has been a long-standing focus of research in the field of aerial mapping. A multi-sensor UAV system, integrating the Global Positioning System (GPS), Inertial Measurement Unit (IMU), 4D millimeter-wave radar and camera, can provide an effective solution to this problem. In this paper, we utilize multi-sensor data to overcome the limitations of conventional orthoimage generation methods in terms of temporal performance, system robustness, and geographic reference accuracy. A prior-pose-optimized feature matching method is introduced to enhance matching speed and accuracy, reducing the number of required features and providing precise references for the Structure from Motion (SfM) process. The proposed method exhibits robustness in low-texture scenes like farmlands, where feature matching is difficult. Experiments show that our approach achieves accurate feature matching orthoimage generation in a short time. The proposed drone system effectively aids in farmland detection and management.
♻ ☆ Q-Eval-100K: Evaluating Visual Quality and Alignment Level for Text-to-Vision Content CVPR 2025
Evaluating text-to-vision content hinges on two crucial aspects: visual quality and alignment. While significant progress has been made in developing objective models to assess these dimensions, the performance of such models heavily relies on the scale and quality of human annotations. According to Scaling Law, increasing the number of human-labeled instances follows a predictable pattern that enhances the performance of evaluation models. Therefore, we introduce a comprehensive dataset designed to Evaluate Visual quality and Alignment Level for text-to-vision content (Q-EVAL-100K), featuring the largest collection of human-labeled Mean Opinion Scores (MOS) for the mentioned two aspects. The Q-EVAL-100K dataset encompasses both text-to-image and text-to-video models, with 960K human annotations specifically focused on visual quality and alignment for 100K instances (60K images and 40K videos). Leveraging this dataset with context prompt, we propose Q-Eval-Score, a unified model capable of evaluating both visual quality and alignment with special improvements for handling long-text prompt alignment. Experimental results indicate that the proposed Q-Eval-Score achieves superior performance on both visual quality and alignment, with strong generalization capabilities across other benchmarks. These findings highlight the significant value of the Q-EVAL-100K dataset. Data and codes will be available at https://github.com/zzc-1998/Q-Eval.
comment: Accepted to CVPR 2025
♻ ☆ Floorplan-SLAM: A Real-Time, High-Accuracy, and Long-Term Multi-Session Point-Plane SLAM for Efficient Floorplan Reconstruction
Floorplan reconstruction provides structural priors essential for reliable indoor robot navigation and high-level scene understanding. However, existing approaches either require time-consuming offline processing with a complete map, or rely on expensive sensors and substantial computational resources. To address the problems, we propose Floorplan-SLAM, which incorporates floorplan reconstruction tightly into a multi-session SLAM system by seamlessly interacting with plane extraction, pose estimation, and back-end optimization, achieving real-time, high-accuracy, and long-term floorplan reconstruction using only a stereo camera. Specifically, we present a robust plane extraction algorithm that operates in a compact plane parameter space and leverages spatially complementary features to accurately detect planar structures, even in weakly textured scenes. Furthermore, we propose a floorplan reconstruction module tightly coupled with the SLAM system, which uses continuously optimized plane landmarks and poses to formulate and solve a novel optimization problem, thereby enabling real-time incremental floorplan reconstruction. Note that by leveraging the map merging capability of multi-session SLAM, our method supports long-term floorplan reconstruction across multiple sessions without redundant data collection. Experiments on the VECtor and the self-collected datasets indicate that Floorplan-SLAM significantly outperforms state-of-the-art methods in terms of plane extraction robustness, pose estimation accuracy, and floorplan reconstruction fidelity and speed, achieving real-time performance at 25-45 FPS without GPU acceleration, which reduces the floorplan reconstruction time for a 1000 square meters scene from over 10 hours to just 9.44 minutes.
♻ ☆ KiVA: Kid-inspired Visual Analogies for Testing Large Multimodal Models
This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.
comment: 10 pages. Project website: https://ey242.github.io/kiva.github.io/. Benchmark and code: https://github.com/ey242/KiVA
♻ ☆ NVILA: Efficient Frontier Visual Language Models
Visual language models (VLMs) have made significant advances in accuracy in recent years. However, their efficiency has received much less attention. This paper introduces NVILA, a family of open VLMs designed to optimize both efficiency and accuracy. Building on top of VILA, we improve its model architecture by first scaling up the spatial and temporal resolutions, and then compressing visual tokens. This "scale-then-compress" approach enables NVILA to efficiently process high-resolution images and long videos. We also conduct a systematic investigation to enhance the efficiency of NVILA throughout its entire lifecycle, from training and fine-tuning to deployment. NVILA matches or surpasses the accuracy of many leading open and proprietary VLMs across a wide range of image and video benchmarks. At the same time, it reduces training costs by 4.5X, fine-tuning memory usage by 3.4X, pre-filling latency by 1.6-2.2X, and decoding latency by 1.2-2.8X. We will soon make our code and models available to facilitate reproducibility.
♻ ☆ Fractal Calibration for long-tailed object detection CVPR2025
Real-world datasets follow an imbalanced distribution, which poses significant challenges in rare-category object detection. Recent studies tackle this problem by developing re-weighting and re-sampling methods, that utilise the class frequencies of the dataset. However, these techniques focus solely on the frequency statistics and ignore the distribution of the classes in image space, missing important information. In contrast to them, we propose FRActal CALibration (FRACAL): a novel post-calibration method for long-tailed object detection. FRACAL devises a logit adjustment method that utilises the fractal dimension to estimate how uniformly classes are distributed in image space. During inference, it uses the fractal dimension to inversely downweight the probabilities of uniformly spaced class predictions achieving balance in two axes: between frequent and rare categories, and between uniformly spaced and sparsely spaced classes. FRACAL is a post-processing method and it does not require any training, also it can be combined with many off-the-shelf models such as one-stage sigmoid detectors and two-stage instance segmentation models. FRACAL boosts the rare class performance by up to 8.6% and surpasses all previous methods on LVIS dataset, while showing good generalisation to other datasets such as COCO, V3Det and OpenImages. We provide the code at https://github.com/kostas1515/FRACAL.
comment: CVPR2025
♻ ☆ Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation
Many large-scale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show that sparse coding offers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We propose Contrastive Sparse Representation (CSR), a method that sparsifies pre-trained embeddings into a high-dimensional but selectively activated feature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed-often by large margins-while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount. Code is available at https://github.com/neilwen987/CSR_Adaptive_Rep
comment: A novel sparse coding framework designed for learning adaptive representation
♻ ☆ What to align in multimodal contrastive learning? ICLR 2025
Humans perceive the world through multisensory integration, blending the information of different modalities to adapt their behavior. Contrastive learning offers an appealing solution for multimodal self-supervised learning. Indeed, by considering each modality as a different view of the same entity, it learns to align features of different modalities in a shared representation space. However, this approach is intrinsically limited as it only learns shared or redundant information between modalities, while multimodal interactions can arise in other ways. In this work, we introduce CoMM, a Contrastive MultiModal learning strategy that enables the communication between modalities in a single multimodal space. Instead of imposing cross- or intra- modality constraints, we propose to align multimodal representations by maximizing the mutual information between augmented versions of these multimodal features. Our theoretical analysis shows that shared, synergistic and unique terms of information naturally emerge from this formulation, allowing us to estimate multimodal interactions beyond redundancy. We test CoMM both in a controlled and in a series of real-world settings: in the former, we demonstrate that CoMM effectively captures redundant, unique and synergistic information between modalities. In the latter, CoMM learns complex multimodal interactions and achieves state-of-the-art results on the seven multimodal benchmarks. Code is available at https://github.com/Duplums/CoMM
comment: ICLR 2025, 25 pages
♻ ☆ More than Memes: A Multimodal Topic Modeling Approach to Conspiracy Theories on Telegram
To address the increasing prevalence of (audio-)visual data on social media, and to capture the evolving and dynamic nature of this communication, researchers have begun to explore the potential of unsupervised approaches for analyzing multimodal online content. However, existing research often neglects visual content beyond memes, and in addition lacks methods to compare topic models across modalities. Our study addresses these gaps by applying multimodal topic modeling for analyzing conspiracy theories in German-language Telegram channels. We use BERTopic with CLIP for the analysis of textual and visual data in a corpus of ~40, 000 Telegram messages posted in October 2023 in 571 German-language Telegram channels known for disseminating conspiracy theories. Through this dataset, we provide insights into unimodal and multimodal topic models by analyzing symmetry and intersections of topics across modalities. We demonstrate the variety of textual and visual content shared in the channels discovered through the topic modeling, and propose a conceptual framework for the analysis of textual and visual discursive strategies in the communication of conspiracy theories. We apply the framework in a case study of the topic group Israel Gaza.
comment: 12 pages, 10 figures
♻ ☆ Reasoning to Attend: Try to Understand How Token Works CVPR 2025
Current Large Multimodal Models (LMMs) empowered visual grounding typically rely on $\texttt{}$ token as a text prompt to jointly optimize the vision-language model (e.g., LLaVA) and the downstream task-specified model (\eg, SAM). However, we observe that little research has looked into how it works. In this work, we first visualize the similarity maps, which are obtained by computing the semantic similarity between the $\texttt{}$ token and the image token embeddings derived from the last hidden layer in both the LLaVA encoder and SAM decoder. Intriguingly, we have found that a striking consistency holds in terms of activation responses in the similarity map,which reveals that what $\texttt{}$ token contributes to is the semantic similarity within image-text pairs. Specifically, $\texttt{}$ token, a placeholder expanded in text vocabulary, extensively queries among individual tokenized image patches to match the semantics of an object from text to the paired image while the Large Language Models (LLMs) are being fine-tuned. Upon the above findings, we present READ, which facilitates LMMs' resilient $\textbf{REA}$soning capability of where to atten$\textbf{D}$ under the guidance of highly activated points borrowed from similarity maps. Remarkably, READ features an intuitive design, Similarity as Points module (SasP), which can be seamlessly applied to $\texttt{}$-like paradigms in a plug-and-play fashion. Also, extensive experiments have been conducted on the ReasonSeg and RefCOCO(+/g) datasets. To validate whether READ suffers from catastrophic forgetting of previous skills after fine-tuning, we further assess its generation ability on an augmented FP-RefCOCO(+/g) dataset. All codes and models are publicly available at https://github.com/rui-qian/READ.
comment: This work has been accepted to CVPR 2025, please refer to https://github.com/rui-qian/READ
♻ ☆ StdGEN: Semantic-Decomposed 3D Character Generation from Single Images CVPR 2025
We present StdGEN, an innovative pipeline for generating semantically decomposed high-quality 3D characters from single images, enabling broad applications in virtual reality, gaming, and filmmaking, etc. Unlike previous methods which struggle with limited decomposability, unsatisfactory quality, and long optimization times, StdGEN features decomposability, effectiveness and efficiency; i.e., it generates intricately detailed 3D characters with separated semantic components such as the body, clothes, and hair, in three minutes. At the core of StdGEN is our proposed Semantic-aware Large Reconstruction Model (S-LRM), a transformer-based generalizable model that jointly reconstructs geometry, color and semantics from multi-view images in a feed-forward manner. A differentiable multi-layer semantic surface extraction scheme is introduced to acquire meshes from hybrid implicit fields reconstructed by our S-LRM. Additionally, a specialized efficient multi-view diffusion model and an iterative multi-layer surface refinement module are integrated into the pipeline to facilitate high-quality, decomposable 3D character generation. Extensive experiments demonstrate our state-of-the-art performance in 3D anime character generation, surpassing existing baselines by a significant margin in geometry, texture and decomposability. StdGEN offers ready-to-use semantic-decomposed 3D characters and enables flexible customization for a wide range of applications. Project page: https://stdgen.github.io
comment: CVPR 2025. 13 pages, 10 figures
♻ ☆ Unleashing HyDRa: Hybrid Fusion, Depth Consistency and Radar for Unified 3D Perception
Low-cost, vision-centric 3D perception systems for autonomous driving have made significant progress in recent years, narrowing the gap to expensive LiDAR-based methods. The primary challenge in becoming a fully reliable alternative lies in robust depth prediction capabilities, as camera-based systems struggle with long detection ranges and adverse lighting and weather conditions. In this work, we introduce HyDRa, a novel camera-radar fusion architecture for diverse 3D perception tasks. Building upon the principles of dense BEV (Bird's Eye View)-based architectures, HyDRa introduces a hybrid fusion approach to combine the strengths of complementary camera and radar features in two distinct representation spaces. Our Height Association Transformer module leverages radar features already in the perspective view to produce more robust and accurate depth predictions. In the BEV, we refine the initial sparse representation by a Radar-weighted Depth Consistency. HyDRa achieves a new state-of-the-art for camera-radar fusion of 64.2 NDS (+1.8) and 58.4 AMOTA (+1.5) on the public nuScenes dataset. Moreover, our new semantically rich and spatially accurate BEV features can be directly converted into a powerful occupancy representation, beating all previous camera-based methods on the Occ3D benchmark by an impressive 3.7 mIoU. Code and models are available at https://github.com/phi-wol/hydra.
comment: 10 pages, 7 figures, added eval on VoD, added appendix
♻ ☆ On the Utility of Equivariance and Symmetry Breaking in Deep Learning Architectures on Point Clouds
This paper explores the key factors that influence the performance of models working with point clouds, across different tasks of varying geometric complexity. In this work, we explore the trade-offs between flexibility and weight-sharing introduced by equivariant layers, assessing when equivariance boosts or detracts from performance. It is often argued that providing more information as input improves a model's performance. However, if this additional information breaks certain properties, such as $\SE(3)$ equivariance, does it remain beneficial? We identify the key aspects of equivariant and non-equivariant architectures that drive success in different tasks by benchmarking them on segmentation, regression, and generation tasks across multiple datasets with increasing complexity. We observe a positive impact of equivariance, which becomes more pronounced with increasing task complexity, even when strict equivariance is not required.
comment: 19 pages, 4 figures
♻ ☆ Human-in-the-loop Reasoning For Traffic Sign Detection: Collaborative Approach Yolo With Video-llava
Traffic Sign Recognition (TSR) detection is a crucial component of autonomous vehicles. While You Only Look Once (YOLO) is a popular real-time object detection algorithm, factors like training data quality and adverse weather conditions (e.g., heavy rain) can lead to detection failures. These failures can be particularly dangerous when visual similarities between objects exist, such as mistaking a 30 km/h sign for a higher speed limit sign. This paper proposes a method that combines video analysis and reasoning, prompting with a human-in-the-loop guide large vision model to improve YOLOs accuracy in detecting road speed limit signs, especially in semi-real-world conditions. It is hypothesized that the guided prompting and reasoning abilities of Video-LLava can enhance YOLOs traffic sign detection capabilities. This hypothesis is supported by an evaluation based on human-annotated accuracy metrics within a dataset of recorded videos from the CARLA car simulator. The results demonstrate that a collaborative approach combining YOLO with Video-LLava and reasoning can effectively address challenging situations such as heavy rain and overcast conditions that hinder YOLOs detection capabilities.
comment: 10 pages, 6 figures
♻ ☆ Tiny Robotics Dataset and Benchmark for Continual Object Detection
Detecting objects in mobile robotics is crucial for numerous applications, from autonomous navigation to inspection. However, robots often need to operate in different domains from those they were trained in, requiring them to adjust to these changes. Tiny mobile robots, subject to size, power, and computational constraints, encounter even more difficulties in running and adapting these algorithms. Such adaptability, though, is crucial for real-world deployment, where robots must operate effectively in dynamic and unpredictable settings. In this work, we introduce a novel benchmark to evaluate the continual learning capabilities of object detection systems in tiny robotic platforms. Our contributions include: (i) Tiny Robotics Object Detection~(TiROD), a comprehensive dataset collected using the onboard camera of a small mobile robot, designed to test object detectors across various domains and classes; (ii) a benchmark of different continual learning strategies on this dataset using NanoDet, a lightweight object detector. Our results highlight key challenges in developing robust and efficient continual learning strategies for object detectors in tiny robotics.
♻ ☆ Safety Without Semantic Disruptions: Editing-free Safe Image Generation via Context-preserving Dual Latent Reconstruction
Training multimodal generative models on large, uncurated datasets can result in users being exposed to harmful, unsafe and controversial or culturally-inappropriate outputs. While model editing has been proposed to remove or filter undesirable concepts in embedding and latent spaces, it can inadvertently damage learned manifolds, distorting concepts in close semantic proximity. We identify limitations in current model editing techniques, showing that even benign, proximal concepts may become misaligned. To address the need for safe content generation, we leverage safe embeddings and a modified diffusion process with tunable weighted summation in the latent space to generate safer images. Our method preserves global context without compromising the structural integrity of the learned manifolds. We achieve state-of-the-art results on safe image generation benchmarks and offer intuitive control over the level of model safety. We identify trade-offs between safety and censorship, which presents a necessary perspective in the development of ethical AI models. We will release our code. Keywords: Text-to-Image Models, Generative AI, Safety, Reliability, Model Editing
comment: This research is supported by the NISDRG project #20100007, funded by the Australian Government
♻ ☆ VideoWorld: Exploring Knowledge Learning from Unlabeled Videos
This work explores whether a deep generative model can learn complex knowledge solely from visual input, in contrast to the prevalent focus on text-based models like large language models (LLMs). We develop VideoWorld, an auto-regressive video generation model trained on unlabeled video data, and test its knowledge acquisition abilities in video-based Go and robotic control tasks. Our experiments reveal two key findings: (1) video-only training provides sufficient information for learning knowledge, including rules, reasoning and planning capabilities, and (2) the representation of visual change is crucial for knowledge acquisition. To improve both the efficiency and efficacy of this process, we introduce the Latent Dynamics Model (LDM) as a key component of VideoWorld. Remarkably, VideoWorld reaches a 5-dan professional level in the Video-GoBench with just a 300-million-parameter model, without relying on search algorithms or reward mechanisms typical in reinforcement learning. In robotic tasks, VideoWorld effectively learns diverse control operations and generalizes across environments, approaching the performance of oracle models in CALVIN and RLBench. This study opens new avenues for knowledge acquisition from visual data, with all code, data, and models open-sourced for further research.
comment: Code and models are released at: https://maverickren.github.io/VideoWorld.github.io/
♻ ☆ Perceptual Multi-Exposure Fusion
As an ever-increasing demand for high dynamic range (HDR) scene shooting, multi-exposure image fusion (MEF) technology has abounded. In recent years, multi-scale exposure fusion approaches based on detail-enhancement have led the way for improvement in highlight and shadow details. Most of such methods, however, are too computationally expensive to be deployed on mobile devices. This paper presents a perceptual multi-exposure fusion method that not just ensures fine shadow/highlight details but with lower complexity than detailenhanced methods. We analyze the potential defects of three classical exposure measures in lieu of using detail-enhancement component and improve two of them, namely adaptive Wellexposedness (AWE) and the gradient of color images (3-D gradient). AWE designed in YCbCr color space considers the difference between varying exposure images. 3-D gradient is employed to extract fine details. We build a large-scale multiexposure benchmark dataset suitable for static scenes, which contains 167 image sequences all told. Experiments on the constructed dataset demonstrate that the proposed method exceeds existing eight state-of-the-art approaches in terms of visually and MEF-SSIM value. Moreover, our approach can achieve a better improvement for current image enhancement techniques, ensuring fine detail in bright light.
♻ ☆ DFREC: DeepFake Identity Recovery Based on Identity-aware Masked Autoencoder
Recent advances in deepfake forensics have primarily focused on improving the classification accuracy and generalization performance. Despite enormous progress in detection accuracy across a wide variety of forgery algorithms, existing algorithms lack intuitive interpretability and identity traceability to help with forensic investigation. In this paper, we introduce a novel DeepFake Identity Recovery scheme (DFREC) to fill this gap. DFREC aims to recover the pair of source and target faces from a deepfake image to facilitate deepfake identity tracing and reduce the risk of deepfake attack. It comprises three key components: an Identity Segmentation Module (ISM), a Source Identity Reconstruction Module (SIRM), and a Target Identity Reconstruction Module (TIRM). The ISM segments the input face into distinct source and target face information, and the SIRM reconstructs the source face and extracts latent target identity features with the segmented source information. The background context and latent target identity features are synergetically fused by a Masked Autoencoder in the TIRM to reconstruct the target face. We evaluate DFREC on six different high-fidelity face-swapping attacks on FaceForensics++, CelebaMegaFS and FFHQ-E4S datasets, which demonstrate its superior recovery performance over state-of-the-art deepfake recovery algorithms. In addition, DFREC is the only scheme that can recover both pristine source and target faces directly from the forgery image with high fadelity.
♻ ☆ Deblur-Avatar: Animatable Avatars from Motion-Blurred Monocular Videos
We introduce a novel framework for modeling high-fidelity, animatable 3D human avatars from motion-blurred monocular video inputs. Motion blur is prevalent in real-world dynamic video capture, especially due to human movements in 3D human avatar modeling. Existing methods either (1) assume sharp image inputs, failing to address the detail loss introduced by motion blur, or (2) mainly consider blur by camera movements, neglecting the human motion blur which is more common in animatable avatars. Our proposed approach integrates a human movement-based motion blur model into 3D Gaussian Splatting (3DGS). By explicitly modeling human motion trajectories during exposure time, we jointly optimize the trajectories and 3D Gaussians to reconstruct sharp, high-quality human avatars. We employ a pose-dependent fusion mechanism to distinguish moving body regions, optimizing both blurred and sharp areas effectively. Extensive experiments on synthetic and real-world datasets demonstrate that our method significantly outperforms existing methods in rendering quality and quantitative metrics, producing sharp avatar reconstructions and enabling real-time rendering under challenging motion blur conditions.
♻ ☆ BHViT: Binarized Hybrid Vision Transformer CVPR2025
Model binarization has made significant progress in enabling real-time and energy-efficient computation for convolutional neural networks (CNN), offering a potential solution to the deployment challenges faced by Vision Transformers (ViTs) on edge devices. However, due to the structural differences between CNN and Transformer architectures, simply applying binary CNN strategies to the ViT models will lead to a significant performance drop. To tackle this challenge, we propose BHViT, a binarization-friendly hybrid ViT architecture and its full binarization model with the guidance of three important observations. Initially, BHViT utilizes the local information interaction and hierarchical feature aggregation technique from coarse to fine levels to address redundant computations stemming from excessive tokens. Then, a novel module based on shift operations is proposed to enhance the performance of the binary Multilayer Perceptron (MLP) module without significantly increasing computational overhead. In addition, an innovative attention matrix binarization method based on quantization decomposition is proposed to evaluate the token's importance in the binarized attention matrix. Finally, we propose a regularization loss to address the inadequate optimization caused by the incompatibility between the weight oscillation in the binary layers and the Adam Optimizer. Extensive experimental results demonstrate that our proposed algorithm achieves SOTA performance among binary ViT methods.
comment: Accepted by CVPR2025
♻ ☆ LDPM: Towards undersampled MRI reconstruction with MR-VAE and Latent Diffusion Prior
Diffusion models, as powerful generative models, have found a wide range of applications and shown great potential in solving image reconstruction problems. Some works attempted to solve MRI reconstruction with diffusion models, but these methods operate directly in pixel space, leading to higher computational costs for optimization and inference. Latent diffusion models, pre-trained on natural images with rich visual priors, are expected to solve the high computational cost problem in MRI reconstruction by operating in a lower-dimensional latent space. However, direct application to MRI reconstruction faces three key challenges: (1) absence of explicit control mechanisms for medical fidelity, (2) domain gap between natural images and MR physics, and (3) undefined data consistency in latent space. To address these challenges, a novel Latent Diffusion Prior-based undersampled MRI reconstruction (LDPM) method is proposed. Our LDPM framework addresses these challenges by: (1) a sketch-guided pipeline with a two-step reconstruction strategy, which balances perceptual quality and anatomical fidelity, (2) an MRI-optimized VAE (MR-VAE), which achieves an improvement of approximately 3.92 dB in PSNR for undersampled MRI reconstruction compared to that with SD-VAE \cite{sd}, and (3) Dual-Stage Sampler, a modified version of spaced DDPM sampler, which enforces high-fidelity reconstruction in the latent space. Experiments on the fastMRI dataset\cite{fastmri} demonstrate the state-of-the-art performance of the proposed method and its robustness across various scenarios. The effectiveness of each module is also verified through ablation experiments.
♻ ☆ GSplatLoc: Grounding Keypoint Descriptors into 3D Gaussian Splatting for Improved Visual Localization
Although various visual localization approaches exist, such as scene coordinate regression and camera pose regression, these methods often struggle with optimization complexity or limited accuracy. To address these challenges, we explore the use of novel view synthesis techniques, particularly 3D Gaussian Splatting (3DGS), which enables the compact encoding of both 3D geometry and scene appearance. We propose a two-stage procedure that integrates dense and robust keypoint descriptors from the lightweight XFeat feature extractor into 3DGS, enhancing performance in both indoor and outdoor environments. The coarse pose estimates are directly obtained via 2D-3D correspondences between the 3DGS representation and query image descriptors. In the second stage, the initial pose estimate is refined by minimizing the rendering-based photometric warp loss. Benchmarking on widely used indoor and outdoor datasets demonstrates improvements over recent neural rendering-based localization methods, such as NeRFMatch and PNeRFLoc.
comment: Project website at https://gsplatloc.github.io/
♻ ☆ ArtNVG: Content-Style Separated Artistic Neighboring-View Gaussian Stylization
As demand from the film and gaming industries for 3D scenes with target styles grows, the importance of advanced 3D stylization techniques increases. However, recent methods often struggle to maintain local consistency in color and texture throughout stylized scenes, which is essential for maintaining aesthetic coherence. To solve this problem, this paper introduces ArtNVG, an innovative 3D stylization framework that efficiently generates stylized 3D scenes by leveraging reference style images. Built on 3D Gaussian Splatting (3DGS), ArtNVG achieves rapid optimization and rendering while upholding high reconstruction quality. Our framework realizes high-quality 3D stylization by incorporating two pivotal techniques: Content-Style Separated Control and Attention-based Neighboring-View Alignment. Content-Style Separated Control uses the CSGO model and the Tile ControlNet to decouple the content and style control, reducing risks of information leakage. Concurrently, Attention-based Neighboring-View Alignment ensures consistency of local colors and textures across neighboring views, significantly improving visual quality. Extensive experiments validate that ArtNVG surpasses existing methods, delivering superior results in content preservation, style alignment, and local consistency.
♻ ☆ Multimodal Action Quality Assessment
Action quality assessment (AQA) is to assess how well an action is performed. Previous works perform modelling by only the use of visual information, ignoring audio information. We argue that although AQA is highly dependent on visual information, the audio is useful complementary information for improving the score regression accuracy, especially for sports with background music, such as figure skating and rhythmic gymnastics. To leverage multimodal information for AQA, i.e., RGB, optical flow and audio information, we propose a Progressive Adaptive Multimodal Fusion Network (PAMFN) that separately models modality-specific information and mixed-modality information. Our model consists of with three modality-specific branches that independently explore modality-specific information and a mixed-modality branch that progressively aggregates the modality-specific information from the modality-specific branches. To build the bridge between modality-specific branches and the mixed-modality branch, three novel modules are proposed. First, a Modality-specific Feature Decoder module is designed to selectively transfer modality-specific information to the mixed-modality branch. Second, when exploring the interaction between modality-specific information, we argue that using an invariant multimodal fusion policy may lead to suboptimal results, so as to take the potential diversity in different parts of an action into consideration. Therefore, an Adaptive Fusion Module is proposed to learn adaptive multimodal fusion policies in different parts of an action. This module consists of several FusionNets for exploring different multimodal fusion strategies and a PolicyNet for deciding which FusionNets are enabled. Third, a module called Cross-modal Feature Decoder is designed to transfer cross-modal features generated by Adaptive Fusion Module to the mixed-modality branch.
comment: IEEE Transactions on Image Processing 2024
♻ ☆ MVP-Shot: Multi-Velocity Progressive-Alignment Framework for Few-Shot Action Recognition
Recent few-shot action recognition (FSAR) methods typically perform semantic matching on learned discriminative features to achieve promising performance. However, most FSAR methods focus on single-scale (e.g., frame-level, segment-level, etc) feature alignment, which ignores that human actions with the same semantic may appear at different velocities. To this end, we develop a novel Multi-Velocity Progressive-alignment (MVP-Shot) framework to progressively learn and align semantic-related action features at multi-velocity levels. Concretely, a Multi-Velocity Feature Alignment (MVFA) module is designed to measure the similarity between features from support and query videos with different velocity scales and then merge all similarity scores in a residual fashion. To avoid the multiple velocity features deviating from the underlying motion semantic, our proposed Progressive Semantic-Tailored Interaction (PSTI) module injects velocity-tailored text information into the video feature via feature interaction on channel and temporal domains at different velocities. The above two modules compensate for each other to make more accurate query sample predictions under the few-shot settings. Experimental results show our method outperforms current state-of-the-art methods on multiple standard few-shot benchmarks (i.e., HMDB51, UCF101, Kinetics, and SSv2-small).
comment: Accepted to TMM 2025
♻ ☆ ChartX & ChartVLM: A Versatile Benchmark and Foundation Model for Complicated Chart Reasoning
Recently, many versatile Multi-modal Large Language Models (MLLMs) have emerged continuously. However, their capacity to query information depicted in visual charts and engage in reasoning based on the queried contents remains under-explored. In this paper, to comprehensively and rigorously benchmark the ability of the off-the-shelf MLLMs in the chart domain, we construct ChartX, a multi-modal evaluation set covering 18 chart types, 7 chart tasks, 22 disciplinary topics, and high-quality chart data. Besides, we develop ChartVLM to offer a new perspective on handling multi-modal tasks that strongly depend on interpretable patterns, such as reasoning tasks in the field of charts or geometric images. We evaluate the chart-related ability of mainstream MLLMs and our ChartVLM on the proposed ChartX evaluation set. Extensive experiments demonstrate that ChartVLM surpasses both versatile and chart-related large models, achieving results comparable to GPT-4V. We believe that our study can pave the way for further exploration in creating a more comprehensive chart evaluation set and developing more interpretable multi-modal models. Both ChartX and ChartVLM are available at: https://github.com/Alpha-Innovator/ChartVLM
comment: Code and dataset are available for downloading at: https://github.com/Alpha-Innovator/ChartVLM 26 pages, 15 figures
♻ ☆ LAPTOP-Diff: Layer Pruning and Normalized Distillation for Compressing Diffusion Models
In the era of AIGC, the demand for low-budget or even on-device applications of diffusion models emerged. In terms of compressing the Stable Diffusion models (SDMs), several approaches have been proposed, and most of them leveraged the handcrafted layer removal methods to obtain smaller U-Nets, along with knowledge distillation to recover the network performance. However, such a handcrafting manner of layer removal is inefficient and lacks scalability and generalization, and the feature distillation employed in the retraining phase faces an imbalance issue that a few numerically significant feature loss terms dominate over others throughout the retraining process. To this end, we proposed the layer pruning and normalized distillation for compressing diffusion models (LAPTOP-Diff). We, 1) introduced the layer pruning method to compress SDM's U-Net automatically and proposed an effective one-shot pruning criterion whose one-shot performance is guaranteed by its good additivity property, surpassing other layer pruning and handcrafted layer removal methods, 2) proposed the normalized feature distillation for retraining, alleviated the imbalance issue. Using the proposed LAPTOP-Diff, we compressed the U-Nets of SDXL and SDM-v1.5 for the most advanced performance, achieving a minimal 4.0% decline in PickScore at a pruning ratio of 50% while the comparative methods' minimal PickScore decline is 8.2%.
♻ ☆ Handling Spatial-Temporal Data Heterogeneity for Federated Continual Learning via Tail Anchor CVPR 2025
Federated continual learning (FCL) allows each client to continually update its knowledge from task streams, enhancing the applicability of federated learning in real-world scenarios. However, FCL needs to address not only spatial data heterogeneity between clients but also temporal data heterogeneity between tasks. In this paper, empirical experiments demonstrate that such input-level heterogeneity significantly affects the model's internal parameters and outputs, leading to severe spatial-temporal catastrophic forgetting of local and previous knowledge. To this end, we propose Federated Tail Anchor (FedTA) to mix trainable Tail Anchor with the frozen output features to adjust their position in the feature space, thereby overcoming parameter-forgetting and output-forgetting. Three novel components are also included: Input Enhancement for improving the performance of pre-trained models on downstream tasks; Selective Input Knowledge Fusion for fusion of heterogeneous local knowledge on the server; and Best Global Prototype Selection for finding the best anchor point for each class in the feature space. Extensive experiments demonstrate that FedTA not only outperforms existing FCL methods but also effectively preserves the relative positions of features.
comment: This paper is accepted by CVPR 2025
♻ ☆ Sim2Real within 5 Minutes: Efficient Domain Transfer with Stylized Gaussian Splatting for Endoscopic Images ICRA 2025
Robot assisted endoluminal intervention is an emerging technique for both benign and malignant luminal lesions. With vision-based navigation, when combined with pre-operative imaging data as priors, it is possible to recover position and pose of the endoscope without the need of additional sensors. In practice, however, aligning pre-operative and intra-operative domains is complicated by significant texture differences. Although methods such as style transfer can be used to address this issue, they require large datasets from both source and target domains with prolonged training times. This paper proposes an efficient domain transfer method based on stylized Gaussian splatting, only requiring a few of real images (10 images) with very fast training time. Specifically, the transfer process includes two phases. In the first phase, the 3D models reconstructed from CT scans are represented as differential Gaussian point clouds. In the second phase, only color appearance related parameters are optimized to transfer the style and preserve the visual content. A novel structure consistency loss is applied to latent features and depth levels to enhance the stability of the transferred images. Detailed validation was performed to demonstrate the performance advantages of the proposed method compared to that of the current state-of-the-art, highlighting the potential for intra-operative surgical navigation.
comment: Accepted by ICRA 2025
♻ ☆ A Physical Coherence Benchmark for Evaluating Video Generation Models via Optical Flow-guided Frame Prediction
Recent advances in video generation models demonstrate their potential as world simulators, but they often struggle with videos deviating from physical laws, a key concern overlooked by most text-to-video benchmarks. We introduce a benchmark designed specifically to assess the Physical Coherence of generated videos, PhyCoBench. Our benchmark includes 120 prompts covering 7 categories of physical principles, capturing key physical laws observable in video content. We evaluated four state-of-the-art (SoTA) T2V models on PhyCoBench and conducted manual assessments. Additionally, we propose an automated evaluation model: PhyCoPredictor, a diffusion model that generates optical flow and video frames in a cascade manner. Through a consistency evaluation comparing automated and manual sorting, the experimental results show that PhyCoPredictor currently aligns most closely with human evaluation. Therefore, it can effectively evaluate the physical coherence of videos, providing insights for future model optimization. Our benchmark, including physical coherence prompts, the automatic evaluation tool PhyCoPredictor, and the generated video dataset, has been released on GitHub at https://github.com/Jeckinchen/PhyCoBench.
♻ ☆ Counting Guidance for High Fidelity Text-to-Image Synthesis WACV 2025
Recently, there have been significant improvements in the quality and performance of text-to-image generation, largely due to the impressive results attained by diffusion models. However, text-to-image diffusion models sometimes struggle to create high-fidelity content for the given input prompt. One specific issue is their difficulty in generating the precise number of objects specified in the text prompt. For example, when provided with the prompt "five apples and ten lemons on a table," images generated by diffusion models often contain an incorrect number of objects. In this paper, we present a method to improve diffusion models so that they accurately produce the correct object count based on the input prompt. We adopt a counting network that performs reference-less class-agnostic counting for any given image. We calculate the gradients of the counting network and refine the predicted noise for each step. To address the presence of multiple types of objects in the prompt, we utilize novel attention map guidance to obtain high-quality masks for each object. Finally, we guide the denoising process using the calculated gradients for each object. Through extensive experiments and evaluation, we demonstrate that the proposed method significantly enhances the fidelity of diffusion models with respect to object count. Code is available at https://github.com/furiosa-ai/counting-guidance.
comment: Accepted at WACV 2025 (Oral). Code is available at https://github.com/furiosa-ai/counting-guidance
♻ ☆ HunyuanVideo: A Systematic Framework For Large Video Generative Models
Recent advancements in video generation have significantly impacted daily life for both individuals and industries. However, the leading video generation models remain closed-source, resulting in a notable performance gap between industry capabilities and those available to the public. In this report, we introduce HunyuanVideo, an innovative open-source video foundation model that demonstrates performance in video generation comparable to, or even surpassing, that of leading closed-source models. HunyuanVideo encompasses a comprehensive framework that integrates several key elements, including data curation, advanced architectural design, progressive model scaling and training, and an efficient infrastructure tailored for large-scale model training and inference. As a result, we successfully trained a video generative model with over 13 billion parameters, making it the largest among all open-source models. We conducted extensive experiments and implemented a series of targeted designs to ensure high visual quality, motion dynamics, text-video alignment, and advanced filming techniques. According to evaluations by professionals, HunyuanVideo outperforms previous state-of-the-art models, including Runway Gen-3, Luma 1.6, and three top-performing Chinese video generative models. By releasing the code for the foundation model and its applications, we aim to bridge the gap between closed-source and open-source communities. This initiative will empower individuals within the community to experiment with their ideas, fostering a more dynamic and vibrant video generation ecosystem. The code is publicly available at https://github.com/Tencent/HunyuanVideo.
♻ ☆ SCott: Accelerating Diffusion Models with Stochastic Consistency Distillation
The iterative sampling procedure employed by diffusion models (DMs) often leads to significant inference latency. To address this, we propose Stochastic Consistency Distillation (SCott) to enable accelerated text-to-image generation, where high-quality and diverse generations can be achieved within just 2-4 sampling steps. In contrast to vanilla consistency distillation (CD) which distills the ordinary differential equation solvers-based sampling process of a pre-trained teacher model into a student, SCott explores the possibility and validates the efficacy of integrating stochastic differential equation (SDE) solvers into CD to fully unleash the potential of the teacher. SCott is augmented with elaborate strategies to control the noise strength and sampling process of the SDE solver. An adversarial loss is further incorporated to strengthen the consistency constraints in rare sampling steps. Empirically, on the MSCOCO-2017 5K dataset with a Stable Diffusion-V1.5 teacher, SCott achieves an FID of 21.9 with 2 sampling steps, surpassing that of the 1-step InstaFlow (23.4) and the 4-step UFOGen (22.1). Moreover, SCott can yield more diverse samples than other consistency models for high-resolution image generation, with up to 16% improvement in a qualified metric.
comment: 22 pages, 16 figures
♻ ☆ Schedule On the Fly: Diffusion Time Prediction for Faster and Better Image Generation
Diffusion and flow matching models have achieved remarkable success in text-to-image generation. However, these models typically rely on the predetermined denoising schedules for all prompts. The multi-step reverse diffusion process can be regarded as a kind of chain-of-thought for generating high-quality images step by step. Therefore, diffusion models should reason for each instance to adaptively determine the optimal noise schedule, achieving high generation quality with sampling efficiency. In this paper, we introduce the Time Prediction Diffusion Model (TPDM) for this. TPDM employs a plug-and-play Time Prediction Module (TPM) that predicts the next noise level based on current latent features at each denoising step. We train the TPM using reinforcement learning to maximize a reward that encourages high final image quality while penalizing excessive denoising steps. With such an adaptive scheduler, TPDM not only generates high-quality images that are aligned closely with human preferences but also adjusts diffusion time and the number of denoising steps on the fly, enhancing both performance and efficiency. With Stable Diffusion 3 Medium architecture, TPDM achieves an aesthetic score of 5.44 and a human preference score (HPS) of 29.59, while using around 50% fewer denoising steps to achieve better performance.
♻ ☆ Explaining Vision-Language Similarities in Dual Encoders with Feature-Pair Attributions
Dual encoder architectures like CLIP models map two types of inputs into a shared embedding space and predict similarities between them. Despite their success, it is, however, not understood how these models compare their two inputs. Common first-order feature-attribution methods can only provide limited insights into dual-encoders since their predictions depend on feature-interactions rather than on individual features. In this paper, we first derive a second-order method enabling the attribution of predictions by any differentiable dual encoder onto feature-interactions between its inputs. Second, we apply our method to CLIP models and show that they learn fine-grained correspondences between parts of captions and regions in images. They match objects across input modes also account for mismatches. This visual-linguistic grounding ability, however, varies heavily between object classes and exhibits pronounced out-of-domain effects. We can identify individual errors as well as systematic failure categories including object coverage, unusual scenes and correlated contexts.
♻ ☆ Super-Resolution on Rotationally Scanned Photoacoustic Microscopy Images Incorporating Scanning Prior
Photoacoustic Microscopy (PAM) images integrating the advantages of optical contrast and acoustic resolution have been widely used in brain studies. However, there exists a trade-off between scanning speed and image resolution. Compared with traditional raster scanning, rotational scanning provides good opportunities for fast PAM imaging by optimizing the scanning mechanism. Recently, there is a trend to incorporate deep learning into the scanning process to further increase the scanning speed.Yet, most such attempts are performed for raster scanning while those for rotational scanning are relatively rare. In this study, we propose a novel and well-performing super-resolution framework for rotational scanning-based PAM imaging. To eliminate adjacent rows' displacements due to subject motion or high-frequency scanning distortion,we introduce a registration module across odd and even rows in the preprocessing and incorporate displacement degradation in the training. Besides, gradient-based patch selection is proposed to increase the probability of blood vessel patches being selected for training. A Transformer-based network with a global receptive field is applied for better performance. Experimental results on both synthetic and real datasets demonstrate the effectiveness and generalizability of our proposed framework for rotationally scanned PAM images'super-resolution, both quantitatively and qualitatively. Code is available at https://github.com/11710615/PAMSR.git.
♻ ☆ XLSTM-HVED: Cross-Modal Brain Tumor Segmentation and MRI Reconstruction Method Using Vision XLSTM and Heteromodal Variational Encoder-Decoder
Neurogliomas are among the most aggressive forms of cancer, presenting considerable challenges in both treatment and monitoring due to their unpredictable biological behavior. Magnetic resonance imaging (MRI) is currently the preferred method for diagnosing and monitoring gliomas. However, the lack of specific imaging techniques often compromises the accuracy of tumor segmentation during the imaging process. To address this issue, we introduce the XLSTM-HVED model. This model integrates a hetero-modal encoder-decoder framework with the Vision XLSTM module to reconstruct missing MRI modalities. By deeply fusing spatial and temporal features, it enhances tumor segmentation performance. The key innovation of our approach is the Self-Attention Variational Encoder (SAVE) module, which improves the integration of modal features. Additionally, it optimizes the interaction of features between segmentation and reconstruction tasks through the Squeeze-Fusion-Excitation Cross Awareness (SFECA) module. Our experiments using the BraTS 2024 dataset demonstrate that our model significantly outperforms existing advanced methods in handling cases where modalities are missing. Our source code is available at https://github.com/Quanato607/XLSTM-HVED.
comment: 5 pages, 2 figures
♻ ☆ 3DGS.zip: A survey on 3D Gaussian Splatting Compression Methods
3D Gaussian Splatting (3DGS) has emerged as a cutting-edge technique for real-time radiance field rendering, offering state-of-the-art performance in terms of both quality and speed. 3DGS models a scene as a collection of three-dimensional Gaussians, with additional attributes optimized to conform to the scene's geometric and visual properties. Despite its advantages in rendering speed and image fidelity, 3DGS is limited by its significant storage and memory demands. These high demands make 3DGS impractical for mobile devices or headsets, reducing its applicability in important areas of computer graphics. To address these challenges and advance the practicality of 3DGS, this survey provides a comprehensive and detailed examination of compression and compaction techniques developed to make 3DGS more efficient. We classify existing methods into two categories: compression, which focuses on reducing file size, and compaction, which aims to minimize the number of Gaussians. Both methods aim to maintain or improve quality, each by minimizing its respective attribute: file size for compression and Gaussian count for compaction. We introduce the basic mathematical concepts underlying the analyzed methods, as well as key implementation details and design choices. Our report thoroughly discusses similarities and differences among the methods, as well as their respective advantages and disadvantages. We establish a consistent framework for comparing the surveyed methods based on key performance metrics and datasets. Specifically, since these methods have been developed in parallel and over a short period of time, currently, no comprehensive comparison exists. This survey, for the first time, presents a unified framework to evaluate 3DGS compression techniques. We maintain a website that will be regularly updated with emerging methods: https://w-m.github.io/3dgs-compression-survey/ .
comment: 3D Gaussian Splatting compression survey; 3DGS compression; updated discussion; new approaches added; new illustrations
♻ ☆ SLTNet: Efficient Event-based Semantic Segmentation with Spike-driven Lightweight Transformer-based Networks IROS 2025
Event-based semantic segmentation has great potential in autonomous driving and robotics due to the advantages of event cameras, such as high dynamic range, low latency, and low power cost. Unfortunately, current artificial neural network (ANN)-based segmentation methods suffer from high computational demands, the requirements for image frames, and massive energy consumption, limiting their efficiency and application on resource-constrained edge/mobile platforms. To address these problems, we introduce SLTNet, a spike-driven lightweight transformer-based network designed for event-based semantic segmentation. Specifically, SLTNet is built on efficient spike-driven convolution blocks (SCBs) to extract rich semantic features while reducing the model's parameters. Then, to enhance the long-range contextural feature interaction, we propose novel spike-driven transformer blocks (STBs) with binary mask operations. Based on these basic blocks, SLTNet employs a high-efficiency single-branch architecture while maintaining the low energy consumption of the Spiking Neural Network (SNN). Finally, extensive experiments on DDD17 and DSEC-Semantic datasets demonstrate that SLTNet outperforms state-of-the-art (SOTA) SNN-based methods by at most 9.06% and 9.39% mIoU, respectively, with extremely 4.58x lower energy consumption and 114 FPS inference speed. Our code is open-sourced and available at https://github.com/longxianlei/SLTNet-v1.0.
comment: Submitted to 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025)
♻ ☆ ChemVLM: Exploring the Power of Multimodal Large Language Models in Chemistry Area
Large Language Models (LLMs) have achieved remarkable success and have been applied across various scientific fields, including chemistry. However, many chemical tasks require the processing of visual information, which cannot be successfully handled by existing chemical LLMs. This brings a growing need for models capable of integrating multimodal information in the chemical domain. In this paper, we introduce \textbf{ChemVLM}, an open-source chemical multimodal large language model specifically designed for chemical applications. ChemVLM is trained on a carefully curated bilingual multimodal dataset that enhances its ability to understand both textual and visual chemical information, including molecular structures, reactions, and chemistry examination questions. We develop three datasets for comprehensive evaluation, tailored to Chemical Optical Character Recognition (OCR), Multimodal Chemical Reasoning (MMCR), and Multimodal Molecule Understanding tasks. We benchmark ChemVLM against a range of open-source and proprietary multimodal large language models on various tasks. Experimental results demonstrate that ChemVLM achieves competitive performance across all evaluated tasks. Our model can be found at https://huggingface.co/AI4Chem/ChemVLM-26B.
comment: 11 pages, updated version
♻ ☆ Scale-Invariant Object Detection by Adaptive Convolution with Unified Global-Local Context
Dense features are important for detecting minute objects in images. Unfortunately, despite the remarkable efficacy of the CNN models in multi-scale object detection, CNN models often fail to detect smaller objects in images due to the loss of dense features during the pooling process. Atrous convolution addresses this issue by applying sparse kernels. However, sparse kernels often can lose the multi-scale detection efficacy of the CNN model. In this paper, we propose an object detection model using a Switchable (adaptive) Atrous Convolutional Network (SAC-Net) based on the efficientDet model. A fixed atrous rate limits the performance of the CNN models in the convolutional layers. To overcome this limitation, we introduce a switchable mechanism that allows for dynamically adjusting the atrous rate during the forward pass. The proposed SAC-Net encapsulates the benefits of both low-level and high-level features to achieve improved performance on multi-scale object detection tasks, without losing the dense features. Further, we apply a depth-wise switchable atrous rate to the proposed network, to improve the scale-invariant features. Finally, we apply global context on the proposed model. Our extensive experiments on benchmark datasets demonstrate that the proposed SAC-Net outperforms the state-of-the-art models by a significant margin in terms of accuracy.
♻ ☆ Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator ICLR 2025
Dataset distillation has emerged as a technique aiming to condense informative features from large, natural datasets into a compact and synthetic form. While recent advancements have refined this technique, its performance is bottlenecked by the prevailing class-specific synthesis paradigm. Under this paradigm, synthetic data is optimized exclusively for a pre-assigned one-hot label, creating an implicit class barrier in feature condensation. This leads to inefficient utilization of the distillation budget and oversight of inter-class feature distributions, which ultimately limits the effectiveness and efficiency, as demonstrated in our analysis. To overcome these constraints, this paper presents the Inter-class Feature Compensator (INFER), an innovative distillation approach that transcends the class-specific data-label framework widely utilized in current dataset distillation methods. Specifically, INFER leverages a Universal Feature Compensator (UFC) to enhance feature integration across classes, enabling the generation of multiple additional synthetic instances from a single UFC input. This significantly improves the efficiency of the distillation budget. Moreover, INFER enriches inter-class interactions during the distillation, thereby enhancing the effectiveness and generalizability of the distilled data. By allowing for the linear interpolation of labels similar to those in the original dataset, INFER meticulously optimizes the synthetic data and dramatically reduces the size of soft labels in the synthetic dataset to almost zero, establishing a new benchmark for efficiency and effectiveness in dataset distillation. In practice, INFER demonstrates state-of-the-art performance across benchmark datasets. For instance, in the ipc = 50 setting on ImageNet-1k with the same compression level, it outperforms SRe2L by 34.5% using ResNet18.
comment: Accepted to ICLR 2025
♻ ☆ Look, Listen, and Answer: Overcoming Biases for Audio-Visual Question Answering NeurIPS 2024
Audio-Visual Question Answering (AVQA) is a complex multi-modal reasoning task, demanding intelligent systems to accurately respond to natural language queries based on audio-video input pairs. Nevertheless, prevalent AVQA approaches are prone to overlearning dataset biases, resulting in poor robustness. Furthermore, current datasets may not provide a precise diagnostic for these methods. To tackle these challenges, firstly, we propose a novel dataset, MUSIC-AVQA-R, crafted in two steps: rephrasing questions within the test split of a public dataset (MUSIC-AVQA) and subsequently introducing distribution shifts to split questions. The former leads to a large, diverse test space, while the latter results in a comprehensive robustness evaluation on rare, frequent, and overall questions. Secondly, we propose a robust architecture that utilizes a multifaceted cycle collaborative debiasing strategy to overcome bias learning. Experimental results show that this architecture achieves state-of-the-art performance on MUSIC-AVQA-R, notably obtaining a significant improvement of 9.32%. Extensive ablation experiments are conducted on the two datasets mentioned to analyze the component effectiveness within the debiasing strategy. Additionally, we highlight the limited robustness of existing multi-modal QA methods through the evaluation on our dataset. We also conduct experiments combining various baselines with our proposed strategy on two datasets to verify its plug-and-play capability. Our dataset and code are available at https://github.com/reml-group/MUSIC-AVQA-R.
comment: Accepted by NeurIPS 2024
♻ ☆ MagicDrive-V2: High-Resolution Long Video Generation for Autonomous Driving with Adaptive Control
The rapid advancement of diffusion models has greatly improved video synthesis, especially in controllable video generation, which is vital for applications like autonomous driving. Although DiT with 3D VAE has become a standard framework for video generation, it introduces challenges in controllable driving video generation, especially for geometry control, rendering existing control methods ineffective. To address these issues, we propose MagicDrive-V2, a novel approach that integrates the MVDiT block and spatial-temporal conditional encoding to enable multi-view video generation and precise geometric control. Additionally, we introduce an efficient method for obtaining contextual descriptions for videos to support diverse textual control, along with a progressive training strategy using mixed video data to enhance training efficiency and generalizability. Consequently, MagicDrive-V2 enables multi-view driving video synthesis with $3.3\times$ resolution and $4\times$ frame count (compared to current SOTA), rich contextual control, and geometric controls. Extensive experiments demonstrate MagicDrive-V2's ability, unlocking broader applications in autonomous driving.
comment: Project Website: https://flymin.github.io/magicdrive-v2/
♻ ☆ TimeRefine: Temporal Grounding with Time Refining Video LLM
Video temporal grounding aims to localize relevant temporal boundaries in a video given a textual prompt. Recent work has focused on enabling Video LLMs to perform video temporal grounding via next-token prediction of temporal timestamps. However, accurately localizing timestamps in videos remains challenging for Video LLMs when relying solely on temporal token prediction. Our proposed TimeRefine addresses this challenge in two ways. First, instead of directly predicting the start and end timestamps, we reformulate the temporal grounding task as a temporal refining task: the model first makes rough predictions and then refines them by predicting offsets to the target segment. This refining process is repeated multiple times, through which the model progressively self-improves its temporal localization accuracy. Second, to enhance the model's temporal perception capabilities, we incorporate an auxiliary prediction head that penalizes the model more if a predicted segment deviates further from the ground truth, thus encouraging the model to make closer and more accurate predictions. Our plug-and-play method can be integrated into most LLM-based temporal grounding approaches. The experimental results demonstrate that TimeRefine achieves 3.6% and 5.0% mIoU improvements on the ActivityNet and Charades-STA datasets, respectively. Code and pretrained models will be released.
♻ ☆ CarPlanner: Consistent Auto-regressive Trajectory Planning for Large-scale Reinforcement Learning in Autonomous Driving CVPR 2025
Trajectory planning is vital for autonomous driving, ensuring safe and efficient navigation in complex environments. While recent learning-based methods, particularly reinforcement learning (RL), have shown promise in specific scenarios, RL planners struggle with training inefficiencies and managing large-scale, real-world driving scenarios. In this paper, we introduce \textbf{CarPlanner}, a \textbf{C}onsistent \textbf{a}uto-\textbf{r}egressive \textbf{Planner} that uses RL to generate multi-modal trajectories. The auto-regressive structure enables efficient large-scale RL training, while the incorporation of consistency ensures stable policy learning by maintaining coherent temporal consistency across time steps. Moreover, CarPlanner employs a generation-selection framework with an expert-guided reward function and an invariant-view module, simplifying RL training and enhancing policy performance. Extensive analysis demonstrates that our proposed RL framework effectively addresses the challenges of training efficiency and performance enhancement, positioning CarPlanner as a promising solution for trajectory planning in autonomous driving. To the best of our knowledge, we are the first to demonstrate that the RL-based planner can surpass both IL- and rule-based state-of-the-arts (SOTAs) on the challenging large-scale real-world dataset nuPlan. Our proposed CarPlanner surpasses RL-, IL-, and rule-based SOTA approaches within this demanding dataset.
comment: CVPR 2025
♻ ☆ Deep Learning-based MRI Reconstruction with Artificial Fourier Transform Network (AFTNet)
Deep complex-valued neural networks (CVNNs) provide a powerful way to leverage complex number operations and representations and have succeeded in several phase-based applications. However, previous networks have not fully explored the impact of complex-valued networks in the frequency domain. Here, we introduce a unified complex-valued deep learning framework-Artificial Fourier Transform Network (AFTNet)-which combines domain-manifold learning and CVNNs. AFTNet can be readily used to solve image inverse problems in domain transformation, especially for accelerated magnetic resonance imaging (MRI) reconstruction and other applications. While conventional methods typically utilize magnitude images or treat the real and imaginary components of k-space data as separate channels, our approach directly processes raw k-space data in the frequency domain, utilizing complex-valued operations. This allows for a mapping between the frequency (k-space) and image domain to be determined through cross-domain learning. We show that AFTNet achieves superior accelerated MRI reconstruction compared to existing approaches. Furthermore, our approach can be applied to various tasks, such as denoised magnetic resonance spectroscopy (MRS) reconstruction and datasets with various contrasts. The AFTNet presented here is a valuable preprocessing component for different preclinical studies and provides an innovative alternative for solving inverse problems in imaging and spectroscopy. The code is available at: https://github.com/yanting-yang/AFT-Net.
♻ ☆ RoboSense: Large-scale Dataset and Benchmark for Egocentric Robot Perception and Navigation in Crowded and Unstructured Environments CVPR2025
Reliable embodied perception from an egocentric perspective is challenging yet essential for autonomous navigation technology of intelligent mobile agents. With the growing demand of social robotics, near-field scene understanding becomes an important research topic in the areas of egocentric perceptual tasks related to navigation in both crowded and unstructured environments. Due to the complexity of environmental conditions and difficulty of surrounding obstacles owing to truncation and occlusion, the perception capability under this circumstance is still inferior. To further enhance the intelligence of mobile robots, in this paper, we setup an egocentric multi-sensor data collection platform based on 3 main types of sensors (Camera, LiDAR and Fisheye), which supports flexible sensor configurations to enable dynamic sight of view from ego-perspective, capturing either near or farther areas. Meanwhile, a large-scale multimodal dataset is constructed, named RoboSense, to facilitate egocentric robot perception. Specifically, RoboSense contains more than 133K synchronized data with 1.4M 3D bounding box and IDs annotated in the full $360^{\circ}$ view, forming 216K trajectories across 7.6K temporal sequences. It has $270\times$ and $18\times$ as many annotations of surrounding obstacles within near ranges as the previous datasets collected for autonomous driving scenarios such as KITTI and nuScenes. Moreover, we define a novel matching criterion for near-field 3D perception and prediction metrics. Based on RoboSense, we formulate 6 popular tasks to facilitate the future research development, where the detailed analysis as well as benchmarks are also provided accordingly. Data desensitization measures have been conducted for privacy protection.
comment: Accepted to CVPR2025
♻ ☆ Dynamic Sparse Training versus Dense Training: The Unexpected Winner in Image Corruption Robustness ICLR 2025
It is generally perceived that Dynamic Sparse Training opens the door to a new era of scalability and efficiency for artificial neural networks at, perhaps, some costs in accuracy performance for the classification task. At the same time, Dense Training is widely accepted as being the "de facto" approach to train artificial neural networks if one would like to maximize their robustness against image corruption. In this paper, we question this general practice. Consequently, we claim that, contrary to what is commonly thought, the Dynamic Sparse Training methods can consistently outperform Dense Training in terms of robustness accuracy, particularly if the efficiency aspect is not considered as a main objective (i.e., sparsity levels between 10% and up to 50%), without adding (or even reducing) resource cost. We validate our claim on two types of data, images and videos, using several traditional and modern deep learning architectures for computer vision and three widely studied Dynamic Sparse Training algorithms. Our findings reveal a new yet-unknown benefit of Dynamic Sparse Training and open new possibilities in improving deep learning robustness beyond the current state of the art.
comment: Accepted at ICLR 2025
♻ ☆ VL-Nav: Real-time Vision-Language Navigation with Spatial Reasoning
Vision-language navigation in unknown environments is crucial for mobile robots. In scenarios such as household assistance and rescue, mobile robots need to understand a human command, such as "find a person wearing black". We present a novel vision-language navigation (VL-Nav) system that integrates efficient spatial reasoning on low-power robots. Unlike prior methods that rely on a single image-level feature similarity to guide a robot, our method integrates pixel-wise vision-language features with curiosity-driven exploration. This approach enables robust navigation to human-instructed instances across diverse environments. We deploy VL-Nav on a four-wheel mobile robot and evaluate its performance through comprehensive navigation tasks in both indoor and outdoor environments, spanning different scales and semantic complexities. Remarkably, VL-Nav operates at a real-time frequency of 30 Hz with a Jetson Orin NX, highlighting its ability to conduct efficient vision-language navigation. Results show that VL-Nav achieves an overall success rate of 86.3%, outperforming previous methods by 44.15%.
♻ ☆ Solving the Catastrophic Forgetting Problem in Generalized Category Discovery CVPR 2024
Generalized Category Discovery (GCD) aims to identify a mix of known and novel categories within unlabeled data sets, providing a more realistic setting for image recognition. Essentially, GCD needs to remember existing patterns thoroughly to recognize novel categories. Recent state-of-the-art method SimGCD transfers the knowledge from known-class data to the learning of novel classes through debiased learning. However, some patterns are catastrophically forgot during adaptation and thus lead to poor performance in novel categories classification. To address this issue, we propose a novel learning approach, LegoGCD, which is seamlessly integrated into previous methods to enhance the discrimination of novel classes while maintaining performance on previously encountered known classes. Specifically, we design two types of techniques termed as Local Entropy Regularization (LER) and Dual-views Kullback Leibler divergence constraint (DKL). The LER optimizes the distribution of potential known class samples in unlabeled data, thus ensuring the preservation of knowledge related to known categories while learning novel classes. Meanwhile, DKL introduces Kullback Leibler divergence to encourage the model to produce a similar prediction distribution of two view samples from the same image. In this way, it successfully avoids mismatched prediction and generates more reliable potential known class samples simultaneously. Extensive experiments validate that the proposed LegoGCD effectively addresses the known category forgetting issue across all datasets, eg, delivering a 7.74% and 2.51% accuracy boost on known and novel classes in CUB, respectively. Our code is available at: https://github.com/Cliffia123/LegoGCD.
comment: Accepted by CVPR 2024
♻ ☆ LiFT: Leveraging Human Feedback for Text-to-Video Model Alignment
Recent advances in text-to-video (T2V) generative models have shown impressive capabilities. However, these models are still inadequate in aligning synthesized videos with human preferences (e.g., accurately reflecting text descriptions), which is particularly difficult to address, as human preferences are subjective and challenging to formalize as objective functions. Existing studies train video quality assessment models that rely on human-annotated ratings for video evaluation but overlook the reasoning behind evaluations, limiting their ability to capture nuanced human criteria. Moreover, aligning T2V model using video-based human feedback remains unexplored. Therefore, this paper proposes LiFT, the first method designed to leverage human feedback for T2V model alignment. Specifically, we first construct a Human Rating Annotation dataset, LiFT-HRA, consisting of approximately 10k human annotations, each including a score and its corresponding rationale. Based on this, we train a reward model LiFT-Critic to learn reward function effectively, which serves as a proxy for human judgment, measuring the alignment between given videos and human expectations. Lastly, we leverage the learned reward function to align the T2V model by maximizing the reward-weighted likelihood. As a case study, we apply our pipeline to CogVideoX-2B, showing that the fine-tuned model outperforms the CogVideoX-5B across all 16 metrics, highlighting the potential of human feedback in improving the alignment and quality of synthesized videos.
comment: Project page: https://codegoat24.github.io/LiFT
♻ ☆ LCV2I: Communication-Efficient and High-Performance Collaborative Perception Framework with Low-Resolution LiDAR
Vehicle-to-Infrastructure (V2I) collaborative perception leverages data collected by infrastructure's sensors to enhance vehicle perceptual capabilities. LiDAR, as a commonly used sensor in cooperative perception, is widely equipped in intelligent vehicles and infrastructure. However, its superior performance comes with a correspondingly high cost. To achieve low-cost V2I, reducing the cost of LiDAR is crucial. Therefore, we study adopting low-resolution LiDAR on the vehicle to minimize cost as much as possible. However, simply reducing the resolution of vehicle's LiDAR results in sparse point clouds, making distant small objects even more blurred. Additionally, traditional communication methods have relatively low bandwidth utilization efficiency. These factors pose challenges for us. To balance cost and perceptual accuracy, we propose a new collaborative perception framework, namely LCV2I. LCV2I uses data collected from cameras and low-resolution LiDAR as input. It also employs feature offset correction modules and regional feature enhancement algorithms to improve feature representation. Finally, we use regional difference map and regional score map to assess the value of collaboration content, thereby improving communication bandwidth efficiency. In summary, our approach achieves high perceptual performance while substantially reducing the demand for high-resolution sensors on the vehicle. To evaluate this algorithm, we conduct 3D object detection in the real-world scenario of DAIR-V2X, demonstrating that the performance of LCV2I consistently surpasses currently existing algorithms.
♻ ☆ Detecting Adversarial Data using Perturbation Forgery CVPR 2025
As a defense strategy against adversarial attacks, adversarial detection aims to identify and filter out adversarial data from the data flow based on discrepancies in distribution and noise patterns between natural and adversarial data. Although previous detection methods achieve high performance in detecting gradient-based adversarial attacks, new attacks based on generative models with imbalanced and anisotropic noise patterns evade detection. Even worse, the significant inference time overhead and limited performance against unseen attacks make existing techniques impractical for real-world use. In this paper, we explore the proximity relationship among adversarial noise distributions and demonstrate the existence of an open covering for these distributions. By training on the open covering of adversarial noise distributions, a detector with strong generalization performance against various types of unseen attacks can be developed. Based on this insight, we heuristically propose Perturbation Forgery, which includes noise distribution perturbation, sparse mask generation, and pseudo-adversarial data production, to train an adversarial detector capable of detecting any unseen gradient-based, generative-based, and physical adversarial attacks. Comprehensive experiments conducted on multiple general and facial datasets, with a wide spectrum of attacks, validate the strong generalization of our method.
comment: Accepted as a conference paper at CVPR 2025
♻ ☆ Uni-Renderer: Unifying Rendering and Inverse Rendering Via Dual Stream Diffusion
Rendering and inverse rendering are pivotal tasks in both computer vision and graphics. The rendering equation is the core of the two tasks, as an ideal conditional distribution transfer function from intrinsic properties to RGB images. Despite achieving promising results of existing rendering methods, they merely approximate the ideal estimation for a specific scene and come with a high computational cost. Additionally, the inverse conditional distribution transfer is intractable due to the inherent ambiguity. To address these challenges, we propose a data-driven method that jointly models rendering and inverse rendering as two conditional generation tasks within a single diffusion framework. Inspired by UniDiffuser, we utilize two distinct time schedules to model both tasks, and with a tailored dual streaming module, we achieve cross-conditioning of two pre-trained diffusion models. This unified approach, named Uni-Renderer, allows the two processes to facilitate each other through a cycle-consistent constrain, mitigating ambiguity by enforcing consistency between intrinsic properties and rendered images. Combined with a meticulously prepared dataset, our method effectively decomposition of intrinsic properties and demonstrates a strong capability to recognize changes during rendering. We will open-source our training and inference code to the public, fostering further research and development in this area.
♻ ☆ Category-level Meta-learned NeRF Priors for Efficient Object Mapping
In 3D object mapping, category-level priors enable efficient object reconstruction and canonical pose estimation, requiring only a single prior per semantic category (e.g., chair, book, laptop). Recently, DeepSDF has predominantly been used as a category-level shape prior, but it struggles to reconstruct sharp geometry and is computationally expensive. In contrast, NeRFs capture fine details but have yet to be effectively integrated with category-level priors in a real-time multi-object mapping framework. To bridge this gap, we introduce PRENOM, a Prior-based Efficient Neural Object Mapper that integrates category-level priors with object-level NeRFs to enhance reconstruction efficiency while enabling canonical object pose estimation. PRENOM gets to know objects on a first-name basis by meta-learning on synthetic reconstruction tasks generated from open-source shape datasets. To account for object category variations, it employs a multi-objective genetic algorithm to optimize the NeRF architecture for each category, balancing reconstruction quality and training time. Additionally, prior-based probabilistic ray sampling directs sampling toward expected object regions, accelerating convergence and improving reconstruction quality under constrained resources. Experimental results on a low-end GPU highlight the ability of PRENOM to achieve high-quality reconstructions while maintaining computational feasibility. Specifically, comparisons with prior-free NeRF-based approaches on a synthetic dataset show a 21% lower Chamfer distance, demonstrating better reconstruction quality. Furthermore, evaluations against other approaches using shape priors on a noisy real-world dataset indicate a 13% improvement averaged across all reconstruction metrics, and comparable pose and size estimation accuracy, while being trained for 5x less time.
♻ ☆ Adapting Pre-Trained Vision Models for Novel Instance Detection and Segmentation DSN
Novel Instance Detection and Segmentation (NIDS) aims at detecting and segmenting novel object instances given a few examples of each instance. We propose a unified, simple, yet effective framework (NIDS-Net) comprising object proposal generation, embedding creation for both instance templates and proposal regions, and embedding matching for instance label assignment. Leveraging recent advancements in large vision methods, we utilize Grounding DINO and Segment Anything Model (SAM) to obtain object proposals with accurate bounding boxes and masks. Central to our approach is the generation of high-quality instance embeddings. We utilized foreground feature averages of patch embeddings from the DINOv2 ViT backbone, followed by refinement through a weight adapter mechanism that we introduce. We show experimentally that our weight adapter can adjust the embeddings locally within their feature space and effectively limit overfitting in the few-shot setting. Furthermore, the weight adapter optimizes weights to enhance the distinctiveness of instance embeddings during similarity computation. This methodology enables a straightforward matching strategy that results in significant performance gains. Our framework surpasses current state-of-the-art methods, demonstrating notable improvements in four detection datasets. In the segmentation tasks on seven core datasets of the BOP challenge, our method outperforms the leading published RGB methods and remains competitive with the best RGB-D method. We have also verified our method using real-world images from a Fetch robot and a RealSense camera. Project Page: https://irvlutd.github.io/NIDSNet/
comment: Project Page: https://irvlutd.github.io/NIDSNet/
♻ ☆ MobileViM: A Light-weight and Dimension-independent Vision Mamba for 3D Medical Image Analysis
Efficient evaluation of three-dimensional (3D) medical images is crucial for diagnostic and therapeutic practices in healthcare. Recent years have seen a substantial uptake in applying deep learning and computer vision to analyse and interpret medical images. Traditional approaches, such as convolutional neural networks (CNNs) and vision transformers (ViTs), face significant computational challenges, prompting the need for architectural advancements. Recent efforts have led to the introduction of novel architectures like the ``Mamba'' model as alternative solutions to traditional CNNs or ViTs. The Mamba model excels in the linear processing of one-dimensional data with low computational demands. However, Mamba's potential for 3D medical image analysis remains underexplored and could face significant computational challenges as the dimension increases. This manuscript presents MobileViM, a streamlined architecture for efficient segmentation of 3D medical images. In the MobileViM network, we invent a new dimension-independent mechanism and a dual-direction traversing approach to incorporate with a vision-Mamba-based framework. MobileViM also features a cross-scale bridging technique to improve efficiency and accuracy across various medical imaging modalities. With these enhancements, MobileViM achieves segmentation speeds exceeding 90 frames per second (FPS) on a single graphics processing unit (i.e., NVIDIA RTX 4090). This performance is over 24 FPS faster than the state-of-the-art deep learning models for processing 3D images with the same computational resources. In addition, experimental evaluations demonstrate that MobileViM delivers superior performance, with Dice similarity scores reaching 92.72%, 86.69%, 80.46%, and 77.43% for PENGWIN, BraTS2024, ATLAS, and Toothfairy2 datasets, respectively, which significantly surpasses existing models.
♻ ☆ Vision-based Geo-Localization of Future Mars Rotorcraft in Challenging Illumination Conditions
Planetary exploration using aerial assets has the potential for unprecedented scientific discoveries on Mars. While NASA's Mars helicopter Ingenuity proved flight in Martian atmosphere is possible, future Mars rotocrafts will require advanced navigation capabilities for long-range flights. One such critical capability is Map-based Localization (MbL) which registers an onboard image to a reference map during flight in order to mitigate cumulative drift from visual odometry. However, significant illumination differences between rotocraft observations and a reference map prove challenging for traditional MbL systems, restricting the operational window of the vehicle. In this work, we investigate a new MbL system and propose Geo-LoFTR, a geometry-aided deep learning model for image registration that is more robust under large illumination differences than prior models. The system is supported by a custom simulation framework that uses real orbital maps to produce large amounts of realistic images of the Martian terrain. Comprehensive evaluations show that our proposed system outperforms prior MbL efforts in terms of localization accuracy under significant lighting and scale variations. Furthermore, we demonstrate the validity of our approach across a simulated Martian day.
♻ ☆ LS-HAR: Language Supervised Human Action Recognition with Salient Fusion, Construction Sites as a Use-Case
Detecting human actions is a crucial task for autonomous robots and vehicles, often requiring the integration of various data modalities for improved accuracy. In this study, we introduce a novel approach to Human Action Recognition (HAR) using language supervision named LS-HAR based on skeleton and visual cues. Our method leverages a language model to guide the feature extraction process in the skeleton encoder. Specifically, we employ learnable prompts for the language model conditioned on the skeleton modality to optimize feature representation. Furthermore, we propose a fusion mechanism that combines dual-modality features using a salient fusion module, incorporating attention and transformer mechanisms to address the modalities' high dimensionality. This fusion process prioritizes informative video frames and body joints, enhancing the recognition accuracy of human actions. Additionally, we introduce a new dataset tailored for real-world robotic applications in construction sites, featuring visual, skeleton, and depth data modalities, named VolvoConstAct. This dataset serves to facilitate the training and evaluation of machine learning models to instruct autonomous construction machines for performing necessary tasks in real-world construction sites. To evaluate our approach, we conduct experiments on our dataset as well as three widely used public datasets: NTU-RGB+D, NTU-RGB+D 120, and NW-UCLA. Results reveal that our proposed method achieves promising performance across all datasets, demonstrating its robustness and potential for various applications. The code, dataset, and demonstration of real-machine experiments are available at: https://mmahdavian.github.io/ls_har/
♻ ☆ CSCPR: Cross-Source-Context Indoor RGB-D Place Recognition
We extend our previous work, PoCo, and present a new algorithm, Cross-Source-Context Place Recognition (CSCPR), for RGB-D indoor place recognition that integrates global retrieval and reranking into an end-to-end model and keeps the consistency of using Context-of-Clusters (CoCs) for feature processing. Unlike prior approaches that primarily focus on the RGB domain for place recognition reranking, CSCPR is designed to handle the RGB-D data. We apply the CoCs to handle cross-sourced and cross-scaled RGB-D point clouds and introduce two novel modules for reranking: the Self-Context Cluster (SCC) and the Cross Source Context Cluster (CSCC), which enhance feature representation and match query-database pairs based on local features, respectively. We also release two new datasets, ScanNetIPR and ARKitIPR. Our experiments demonstrate that CSCPR significantly outperforms state-of-the-art models on these datasets by at least 29.27% in Recall@1 on the ScanNet-PR dataset and 43.24% in the new datasets. Code and datasets will be released.
Information Retrieval 12
☆ Addressing Overprescribing Challenges: Fine-Tuning Large Language Models for Medication Recommendation Tasks
Medication recommendation systems have garnered attention within healthcare for their potential to deliver personalized and efficacious drug combinations based on patient's clinical data. However, existing methodologies encounter challenges in adapting to diverse Electronic Health Records (EHR) systems and effectively utilizing unstructured data, resulting in limited generalization capabilities and suboptimal performance. Recently, interest is growing in harnessing Large Language Models (LLMs) in the medical domain to support healthcare professionals and enhance patient care. Despite the emergence of medical LLMs and their promising results in tasks like medical question answering, their practical applicability in clinical settings, particularly in medication recommendation, often remains underexplored. In this study, we evaluate both general-purpose and medical-specific LLMs for medication recommendation tasks. Our findings reveal that LLMs frequently encounter the challenge of overprescribing, leading to heightened clinical risks and diminished medication recommendation accuracy. To address this issue, we propose Language-Assisted Medication Recommendation (LAMO), which employs a parameter-efficient fine-tuning approach to tailor open-source LLMs for optimal performance in medication recommendation scenarios. LAMO leverages the wealth of clinical information within clinical notes, a resource often underutilized in traditional methodologies. As a result of our approach, LAMO outperforms previous state-of-the-art methods by over 10% in internal validation accuracy. Furthermore, temporal and external validations demonstrate LAMO's robust generalization capabilities across various temporal and hospital contexts. Additionally, an out-of-distribution medication recommendation experiment demonstrates LAMO's remarkable accuracy even with medications outside the training data.
☆ Decoupled Recommender Systems: Exploring Alternative Recommender Ecosystem Designs
Recommender ecosystems are an emerging subject of research. Such research examines how the characteristics of algorithms, recommendation consumers, and item providers influence system dynamics and long-term outcomes. One architectural possibility that has not yet been widely explored in this line of research is the consequences of a configuration in which recommendation algorithms are decoupled from the platforms they serve. This is sometimes called "the friendly neighborhood algorithm store" or "middleware" model. We are particularly interested in how such architectures might offer a range of different distributions of utility across consumers, providers, and recommendation platforms. In this paper, we create a model of a recommendation ecosystem that incorporates algorithm choice and examine the outcomes of such a design.
☆ Intrinsic and Extrinsic Factor Disentanglement for Recommendation in Various Context Scenarios
In recommender systems, the patterns of user behaviors (e.g., purchase, click) may vary greatly in different contexts (e.g., time and location). This is because user behavior is jointly determined by two types of factors: intrinsic factors, which reflect consistent user preference, and extrinsic factors, which reflect external incentives that may vary in different contexts. Differentiating between intrinsic and extrinsic factors helps learn user behaviors better. However, existing studies have only considered differentiating them from a single, pre-defined context (e.g., time or location), ignoring the fact that a user's extrinsic factors may be influenced by the interplay of various contexts at the same time. In this paper, we propose the Intrinsic-Extrinsic Disentangled Recommendation (IEDR) model, a generic framework that differentiates intrinsic from extrinsic factors considering various contexts simultaneously, enabling more accurate differentiation of factors and hence the improvement of recommendation accuracy. IEDR contains a context-invariant contrastive learning component to capture intrinsic factors, and a disentanglement component to extract extrinsic factors under the interplay of various contexts. The two components work together to achieve effective factor learning. Extensive experiments on real-world datasets demonstrate IEDR's effectiveness in learning disentangled factors and significantly improving recommendation accuracy by up to 4% in NDCG.
comment: 32 pages, 13 figures, 11 tables. Accepted by Transactions of Information Systems
☆ A Predict-Then-Optimize Customer Allocation Framework for Online Fund Recommendation DASFAA 2025
With the rapid growth of online investment platforms, funds can be distributed to individual customers online. The central issue is to match funds with potential customers under constraints. Most mainstream platforms adopt the recommendation formulation to tackle the problem. However, the traditional recommendation regime has its inherent drawbacks when applying the fund-matching problem with multiple constraints. In this paper, we model the fund matching under the allocation formulation. We design PTOFA, a Predict-Then-Optimize Fund Allocation framework. This data-driven framework consists of two stages, i.e., prediction and optimization, which aim to predict expected revenue based on customer behavior and optimize the impression allocation to achieve the maximum revenue under the necessary constraints, respectively. Extensive experiments on real-world datasets from an industrial online investment platform validate the effectiveness and efficiency of our solution. Additionally, the online A/B tests demonstrate PTOFA's effectiveness in the real-world fund recommendation scenario.
comment: Accepted by DASFAA 2025
♻ ☆ MAPS: Motivation-Aware Personalized Search via LLM-Driven Consultation Alignment
Personalized product search aims to retrieve and rank items that match users' preferences and search intent. Despite their effectiveness, existing approaches typically assume that users' query fully captures their real motivation. However, our analysis of a real-world e-commerce platform reveals that users often engage in relevant consultations before searching, indicating they refine intents through consultations based on motivation and need. The implied motivation in consultations is a key enhancing factor for personalized search. This unexplored area comes with new challenges including aligning contextual motivations with concise queries, bridging the category-text gap, and filtering noise within sequence history. To address these, we propose a Motivation-Aware Personalized Search (MAPS) method. It embeds queries and consultations into a unified semantic space via LLMs, utilizes a Mixture of Attention Experts (MoAE) to prioritize critical semantics, and introduces dual alignment: (1) contrastive learning aligns consultations, reviews, and product features; (2) bidirectional attention integrates motivation-aware embeddings with user preferences. Extensive experiments on real and synthetic data show MAPS outperforms existing methods in both retrieval and ranking tasks.
comment: added project repository & dataset URL
♻ ☆ Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation
Many large-scale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show that sparse coding offers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We propose Contrastive Sparse Representation (CSR), a method that sparsifies pre-trained embeddings into a high-dimensional but selectively activated feature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed-often by large margins-while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount. Code is available at https://github.com/neilwen987/CSR_Adaptive_Rep
comment: A novel sparse coding framework designed for learning adaptive representation
♻ ☆ Beyond Content Relevance: Evaluating Instruction Following in Retrieval Models
Instruction-following capabilities in LLMs have progressed significantly, enabling more complex user interactions through detailed prompts. However, retrieval systems have not matched these advances, most of them still relies on traditional lexical and semantic matching techniques that fail to fully capture user intent. Recent efforts have introduced instruction-aware retrieval models, but these primarily focus on intrinsic content relevance, which neglects the importance of customized preferences for broader document-level attributes. This study evaluates the instruction-following capabilities of various retrieval models beyond content relevance, including LLM-based dense retrieval and reranking models. We develop InfoSearch, a novel retrieval evaluation benchmark spanning six document-level attributes: Audience, Keyword, Format, Language, Length, and Source, and introduce novel metrics -- Strict Instruction Compliance Ratio (SICR) and Weighted Instruction Sensitivity Evaluation (WISE) to accurately assess the models' responsiveness to instructions. Our findings indicate that although fine-tuning models on instruction-aware retrieval datasets and increasing model size enhance performance, most models still fall short of instruction compliance.
♻ ☆ Use of Air Quality Sensor Network Data for Real-time Pollution-Aware POI Suggestion
This demo paper introduces AirSense-R, a privacy-preserving mobile application that delivers real-time, pollution-aware recommendations for urban points of interest (POIs). By merging live air quality data from AirSENCE sensor networks in Bari (Italy) and Cork (Ireland) with user preferences, the system enables health-conscious decision-making. It employs collaborative filtering for personalization, federated learning for privacy, and a prediction engine to detect anomalies and interpolate sparse sensor data. The proposed solution adapts dynamically to urban air quality while safeguarding user privacy. The code and demonstration video are available at https://github.com/AirtownApp/Airtown-Application.git.
♻ ☆ MCiteBench: A Benchmark for Multimodal Citation Text Generation in MLLMs
Multimodal Large Language Models (MLLMs) have advanced in integrating diverse modalities but frequently suffer from hallucination. A promising solution to mitigate this issue is to generate text with citations, providing a transparent chain for verification. However, existing work primarily focuses on generating citations for text-only content, overlooking the challenges and opportunities of multimodal contexts. To address this gap, we introduce MCiteBench, the first benchmark designed to evaluate and analyze the multimodal citation text generation ability of MLLMs. Our benchmark comprises data derived from academic papers and review-rebuttal interactions, featuring diverse information sources and multimodal content. We comprehensively evaluate models from multiple dimensions, including citation quality, source reliability, and answer accuracy. Through extensive experiments, we observe that MLLMs struggle with multimodal citation text generation. We also conduct deep analyses of models' performance, revealing that the bottleneck lies in attributing the correct sources rather than understanding the multimodal content.
♻ ☆ Long-Sequence Recommendation Models Need Decoupled Embeddings ICLR 2025
Lifelong user behavior sequences are crucial for capturing user interests and predicting user responses in modern recommendation systems. A two-stage paradigm is typically adopted to handle these long sequences: a subset of relevant behaviors is first searched from the original long sequences via an attention mechanism in the first stage and then aggregated with the target item to construct a discriminative representation for prediction in the second stage. In this work, we identify and characterize, for the first time, a neglected deficiency in existing long-sequence recommendation models: a single set of embeddings struggles with learning both attention and representation, leading to interference between these two processes. Initial attempts to address this issue with some common methods (e.g., linear projections -- a technique borrowed from language processing) proved ineffective, shedding light on the unique challenges of recommendation models. To overcome this, we propose the Decoupled Attention and Representation Embeddings (DARE) model, where two distinct embedding tables are initialized and learned separately to fully decouple attention and representation. Extensive experiments and analysis demonstrate that DARE provides more accurate searches of correlated behaviors and outperforms baselines with AUC gains up to 0.9% on public datasets and notable improvements on Tencent's advertising platform. Furthermore, decoupling embedding spaces allows us to reduce the attention embedding dimension and accelerate the search procedure by 50% without significant performance impact, enabling more efficient, high-performance online serving. Code in PyTorch for experiments, including model analysis, is available at https://github.com/thuml/DARE.
comment: ICLR 2025. First three authors contributed equally. Code is available at https://github.com/thuml/DARE
♻ ☆ OkraLong: A Flexible Retrieval-Augmented Framework for Long-Text Query Processing
Large Language Models (LLMs) encounter challenges in efficiently processing long-text queries, as seen in applications like enterprise document analysis and financial report comprehension. While conventional solutions employ long-context processing or Retrieval-Augmented Generation (RAG), they suffer from prohibitive input expenses or incomplete information. Recent advancements adopt context compression and dynamic retrieval loops, but still sacrifice critical details or incur iterative costs. To address these limitations, we propose OkraLong, a novel framework that flexibly optimizes the entire processing workflow. Unlike prior static or coarse-grained adaptive strategies, OkraLong adopts fine-grained orchestration through three synergistic components: analyzer, organizer and executor. The analyzer characterizes the task states, which guide the organizer in dynamically scheduling the workflow. The executor carries out the execution and generates the final answer. Experimental results demonstrate that OkraLong not only enhances answer accuracy but also achieves cost-effectiveness across a variety of datasets.
♻ ☆ FLARE: Fusing Language Models and Collaborative Architectures for Recommender Enhancement
Recent proposals in recommender systems represent items with their textual description, using a large language model. They show better results on standard benchmarks compared to an item ID-only model, such as Bert4Rec. In this work, we revisit the often-used Bert4Rec baseline and show that with further tuning, Bert4Rec significantly outperforms previously reported numbers, and in some datasets, is competitive with state-of-the-art models. With revised baselines for item ID-only models, this paper also establishes new competitive results for architectures that combine IDs and textual descriptions. We demonstrate this with Flare (Fusing Language models and collaborative Architectures for Recommender Enhancement). Flare is a novel hybrid sequence recommender that integrates a language model with a collaborative filtering model using a Perceiver network. Prior studies focus evaluation on datasets with limited-corpus size, but many commercially-applicable recommender systems common on the web must handle larger corpora. We evaluate Flare on a more realistic dataset with a significantly larger item vocabulary, introducing new baselines for this setting. This paper also showcases Flare's inherent ability to support critiquing, enabling users to provide feedback and refine recommendations. We leverage critiquing as an evaluation method to assess the model's language understanding and its transferability to the recommendation task.
Machine Learning 152
☆ The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
As large language models (LLMs) become more capable and agentic, the requirement for trust in their outputs grows significantly, yet at the same time concerns have been mounting that models may learn to lie in pursuit of their goals. To address these concerns, a body of work has emerged around the notion of "honesty" in LLMs, along with interventions aimed at mitigating deceptive behaviors. However, evaluations of honesty are currently highly limited, with no benchmark combining large scale and applicability to all models. Moreover, many benchmarks claiming to measure honesty in fact simply measure accuracy--the correctness of a model's beliefs--in disguise. In this work, we introduce a large-scale human-collected dataset for measuring honesty directly, allowing us to disentangle accuracy from honesty for the first time. Across a diverse set of LLMs, we find that while larger models obtain higher accuracy on our benchmark, they do not become more honest. Surprisingly, while most frontier LLMs obtain high scores on truthfulness benchmarks, we find a substantial propensity in frontier LLMs to lie when pressured to do so, resulting in low honesty scores on our benchmark. We find that simple methods, such as representation engineering interventions, can improve honesty. These results underscore the growing need for robust evaluations and effective interventions to ensure LLMs remain trustworthy.
comment: Website: https://www.mask-benchmark.ai
☆ PacketCLIP: Multi-Modal Embedding of Network Traffic and Language for Cybersecurity Reasoning
Traffic classification is vital for cybersecurity, yet encrypted traffic poses significant challenges. We present PacketCLIP, a multi-modal framework combining packet data with natural language semantics through contrastive pretraining and hierarchical Graph Neural Network (GNN) reasoning. PacketCLIP integrates semantic reasoning with efficient classification, enabling robust detection of anomalies in encrypted network flows. By aligning textual descriptions with packet behaviors, it offers enhanced interpretability, scalability, and practical applicability across diverse security scenarios. PacketCLIP achieves a 95% mean AUC, outperforms baselines by 11.6%, and reduces model size by 92%, making it ideal for real-time anomaly detection. By bridging advanced machine learning techniques and practical cybersecurity needs, PacketCLIP provides a foundation for scalable, efficient, and interpretable solutions to tackle encrypted traffic classification and network intrusion detection challenges in resource-constrained environments.
comment: 7 pages, 7 figures
☆ Constrained Gaussian Wasserstein Optimal Transport with Commutative Covariance Matrices
Optimal transport has found widespread applications in signal processing and machine learning. Among its many equivalent formulations, optimal transport seeks to reconstruct a random variable/vector with a prescribed distribution at the destination while minimizing the expected distortion relative to a given random variable/vector at the source. However, in practice, certain constraints may render the optimal transport plan infeasible. In this work, we consider three types of constraints: rate constraints, dimension constraints, and channel constraints, motivated by perception-aware lossy compression, generative principal component analysis, and deep joint source-channel coding, respectively. Special attenion is given to the setting termed Gaussian Wasserstein optimal transport, where both the source and reconstruction variables are multivariate Gaussian, and the end-to-end distortion is measured by the mean squared error. We derive explicit results for the minimum achievable mean squared error under the three aforementioned constraints when the covariance matrices of the source and reconstruction variables commute.
☆ Opportunistic Routing in Wireless Communications via Learnable State-Augmented Policies
This paper addresses the challenge of packet-based information routing in large-scale wireless communication networks. The problem is framed as a constrained statistical learning task, where each network node operates using only local information. Opportunistic routing exploits the broadcast nature of wireless communication to dynamically select optimal forwarding nodes, enabling the information to reach the destination through multiple relay nodes simultaneously. To solve this, we propose a State-Augmentation (SA) based distributed optimization approach aimed at maximizing the total information handled by the source nodes in the network. The problem formulation leverages Graph Neural Networks (GNNs), which perform graph convolutions based on the topological connections between network nodes. Using an unsupervised learning paradigm, we extract routing policies from the GNN architecture, enabling optimal decisions for source nodes across various flows. Numerical experiments demonstrate that the proposed method achieves superior performance when training a GNN-parameterized model, particularly when compared to baseline algorithms. Additionally, applying the method to real-world network topologies and wireless ad-hoc network test beds validates its effectiveness, highlighting the robustness and transferability of GNNs.
☆ Towards Understanding Distilled Reasoning Models: A Representational Approach
In this paper, we investigate how model distillation impacts the development of reasoning features in large language models (LLMs). To explore this, we train a crosscoder on Qwen-series models and their fine-tuned variants. Our results suggest that the crosscoder learns features corresponding to various types of reasoning, including self-reflection and computation verification. Moreover, we observe that distilled models contain unique reasoning feature directions, which could be used to steer the model into over-thinking or incisive-thinking mode. In particular, we perform analysis on four specific reasoning categories: (a) self-reflection, (b) deductive reasoning, (c) alternative reasoning, and (d) contrastive reasoning. Finally, we examine the changes in feature geometry resulting from the distillation process and find indications that larger distilled models may develop more structured representations, which correlate with enhanced distillation performance. By providing insights into how distillation modifies the model, our study contributes to enhancing the transparency and reliability of AI systems.
comment: 13 pages, 11 figures
☆ Graph-Augmented LSTM for Forecasting Sparse Anomalies in Graph-Structured Time Series
Detecting anomalies in time series data is a critical task across many domains. The challenge intensifies when anomalies are sparse and the data are multivariate with relational dependencies across sensors or nodes. Traditional univariate anomaly detectors struggle to capture such cross-node dependencies, particularly in sparse anomaly settings. To address this, we propose a graph-augmented time series forecasting approach that explicitly integrates the graph of relationships among time series into an LSTM forecasting model. This enables the model to detect rare anomalies that might otherwise go unnoticed in purely univariate approaches. We evaluate the approach on two benchmark datasets - the Yahoo Webscope S5 anomaly dataset and the METR-LA traffic sensor network - and compare the performance of the Graph-Augmented LSTM against LSTM-only, ARIMA, and Prophet baselines. Results demonstrate that the graph-augmented model achieves significantly higher precision and recall, improving F1-score by up to 10% over the best baseline
comment: 12 pages
☆ Deep Causal Behavioral Policy Learning: Applications to Healthcare
We present a deep learning-based approach to studying dynamic clinical behavioral regimes in diverse non-randomized healthcare settings. Our proposed methodology - deep causal behavioral policy learning (DC-BPL) - uses deep learning algorithms to learn the distribution of high-dimensional clinical action paths, and identifies the causal link between these action paths and patient outcomes. Specifically, our approach: (1) identifies the causal effects of provider assignment on clinical outcomes; (2) learns the distribution of clinical actions a given provider would take given evolving patient information; (3) and combines these steps to identify the optimal provider for a given patient type and emulate that provider's care decisions. Underlying this strategy, we train a large clinical behavioral model (LCBM) on electronic health records data using a transformer architecture, and demonstrate its ability to estimate clinical behavioral policies. We propose a novel interpretation of a behavioral policy learned using the LCBM: that it is an efficient encoding of complex, often implicit, knowledge used to treat a patient. This allows us to learn a space of policies that are critical to a wide range of healthcare applications, in which the vast majority of clinical knowledge is acquired tacitly through years of practice and only a tiny fraction of information relevant to patient care is written down (e.g. in textbooks, studies or standardized guidelines).
☆ Handling Uncertainty in Health Data using Generative Algorithms
Understanding and managing uncertainty is crucial in machine learning, especially in high-stakes domains like healthcare, where class imbalance can impact predictions. This paper introduces RIGA, a novel pipeline that mitigates class imbalance using generative AI. By converting tabular healthcare data into images, RIGA leverages models like cGAN, VQVAE, and VQGAN to generate balanced samples, improving classification performance. These representations are processed by CNNs and later transformed back into tabular format for seamless integration. This approach enhances traditional classifiers like XGBoost, improves Bayesian structure learning, and strengthens ML model robustness by generating realistic synthetic data for underrepresented classes.
☆ Improving LLM Safety Alignment with Dual-Objective Optimization
Existing training-time safety alignment techniques for large language models (LLMs) remain vulnerable to jailbreak attacks. Direct preference optimization (DPO), a widely deployed alignment method, exhibits limitations in both experimental and theoretical contexts as its loss function proves suboptimal for refusal learning. Through gradient-based analysis, we identify these shortcomings and propose an improved safety alignment that disentangles DPO objectives into two components: (1) robust refusal training, which encourages refusal even when partial unsafe generations are produced, and (2) targeted unlearning of harmful knowledge. This approach significantly increases LLM robustness against a wide range of jailbreak attacks, including prefilling, suffix, and multi-turn attacks across both in-distribution and out-of-distribution scenarios. Furthermore, we introduce a method to emphasize critical refusal tokens by incorporating a reward-based token-level weighting mechanism for refusal learning, which further improves the robustness against adversarial exploits. Our research also suggests that robustness to jailbreak attacks is correlated with token distribution shifts in the training process and internal representations of refusal and harmful tokens, offering valuable directions for future research in LLM safety alignment. The code is available at https://github.com/wicai24/DOOR-Alignment
☆ Curating Demonstrations using Online Experience
Many robot demonstration datasets contain heterogeneous demonstrations of varying quality. This heterogeneity may benefit policy pre-training, but can hinder robot performance when used with a final imitation learning objective. In particular, some strategies in the data may be less reliable than others or may be underrepresented in the data, leading to poor performance when such strategies are sampled at test time. Moreover, such unreliable or underrepresented strategies can be difficult even for people to discern, and sifting through demonstration datasets is time-consuming and costly. On the other hand, policy performance when trained on such demonstrations can reflect the reliability of different strategies. We thus propose for robots to self-curate based on online robot experience (Demo-SCORE). More specifically, we train and cross-validate a classifier to discern successful policy roll-outs from unsuccessful ones and use the classifier to filter heterogeneous demonstration datasets. Our experiments in simulation and the real world show that Demo-SCORE can effectively identify suboptimal demonstrations without manual curation. Notably, Demo-SCORE achieves over 15-35% higher absolute success rate in the resulting policy compared to the base policy trained with all original demonstrations.
☆ Effective LLM Knowledge Learning via Model Generalization
Large language models (LLMs) are trained on enormous documents that contain extensive world knowledge. However, it is still not well-understood how knowledge is acquired via autoregressive pre-training. This lack of understanding greatly hinders effective knowledge learning, especially for continued pretraining on up-to-date information, as this evolving information often lacks diverse repetitions like foundational knowledge. In this paper, we focus on understanding and improving LLM knowledge learning. We found and verified that knowledge learning for LLMs can be deemed as an implicit supervised task hidden in the autoregressive pre-training objective. Our findings suggest that knowledge learning for LLMs would benefit from methods designed to improve generalization ability for supervised tasks. Based on our analysis, we propose the formatting-based data augmentation to grow in-distribution samples, which does not present the risk of altering the facts embedded in documents as text paraphrasing. We also introduce sharpness-aware minimization as an effective optimization algorithm to better improve generalization. Moreover, our analysis and method can be readily extended to instruction tuning. Extensive experiment results validate our findings and demonstrate our methods' effectiveness in both continued pre-training and instruction tuning. This paper offers new perspectives and insights to interpret and design effective strategies for LLM knowledge learning.
☆ A Practical Memory Injection Attack against LLM Agents
Agents based on large language models (LLMs) have demonstrated strong capabilities in a wide range of complex, real-world applications. However, LLM agents with a compromised memory bank may easily produce harmful outputs when the past records retrieved for demonstration are malicious. In this paper, we propose a novel Memory INJection Attack, MINJA, that enables the injection of malicious records into the memory bank by only interacting with the agent via queries and output observations. These malicious records are designed to elicit a sequence of malicious reasoning steps leading to undesirable agent actions when executing the victim user's query. Specifically, we introduce a sequence of bridging steps to link the victim query to the malicious reasoning steps. During the injection of the malicious record, we propose an indication prompt to guide the agent to autonomously generate our designed bridging steps. We also propose a progressive shortening strategy that gradually removes the indication prompt, such that the malicious record will be easily retrieved when processing the victim query comes after. Our extensive experiments across diverse agents demonstrate the effectiveness of MINJA in compromising agent memory. With minimal requirements for execution, MINJA enables any user to influence agent memory, highlighting practical risks of LLM agents.
☆ Towards Trustworthy Federated Learning
This paper develops a comprehensive framework to address three critical trustworthy challenges in federated learning (FL): robustness against Byzantine attacks, fairness, and privacy preservation. To improve the system's defense against Byzantine attacks that send malicious information to bias the system's performance, we develop a Two-sided Norm Based Screening (TNBS) mechanism, which allows the central server to crop the gradients that have the l lowest norms and h highest norms. TNBS functions as a screening tool to filter out potential malicious participants whose gradients are far from the honest ones. To promote egalitarian fairness, we adopt the q-fair federated learning (q-FFL). Furthermore, we adopt a differential privacy-based scheme to prevent raw data at local clients from being inferred by curious parties. Convergence guarantees are provided for the proposed framework under different scenarios. Experimental results on real datasets demonstrate that the proposed framework effectively improves robustness and fairness while managing the trade-off between privacy and accuracy. This work appears to be the first study that experimentally and theoretically addresses fairness, privacy, and robustness in trustworthy FL.
☆ Optimally Installing Strict Equilibria
In this work, we develop a reward design framework for installing a desired behavior as a strict equilibrium across standard solution concepts: dominant strategy equilibrium, Nash equilibrium, correlated equilibrium, and coarse correlated equilibrium. We also extend our framework to capture the Markov-perfect equivalents of each solution concept. Central to our framework is a comprehensive mathematical characterization of strictly installable, based on the desired solution concept and the behavior's structure. These characterizations lead to efficient iterative algorithms, which we generalize to handle optimization objectives through linear programming. Finally, we explore how our results generalize to bounded rational agents.
☆ Analogical Reasoning Inside Large Language Models: Concept Vectors and the Limits of Abstraction
Analogical reasoning relies on conceptual abstractions, but it is unclear whether Large Language Models (LLMs) harbor such internal representations. We explore distilled representations from LLM activations and find that function vectors (FVs; Todd et al., 2024) - compact representations for in-context learning (ICL) tasks - are not invariant to simple input changes (e.g., open-ended vs. multiple-choice), suggesting they capture more than pure concepts. Using representational similarity analysis (RSA), we localize a small set of attention heads that encode invariant concept vectors (CVs) for verbal concepts like "antonym". These CVs function as feature detectors that operate independently of the final output - meaning that a model may form a correct internal representation yet still produce an incorrect output. Furthermore, CVs can be used to causally guide model behaviour. However, for more abstract concepts like "previous" and "next", we do not observe invariant linear representations, a finding we link to generalizability issues LLMs display within these domains.
comment: Preprint
☆ Chunking the Critic: A Transformer-based Soft Actor-Critic with N-Step Returns
Soft Actor-Critic (SAC) critically depends on its critic network, which typically evaluates a single state-action pair to guide policy updates. Using N-step returns is a common practice to reduce the bias in the target values of the critic. However, using N-step returns can again introduce high variance and necessitates importance sampling, often destabilizing training. Recent algorithms have also explored action chunking-such as direct action repetition and movement primitives-to enhance exploration. In this paper, we propose a Transformer-based Critic Network for SAC that integrates the N-returns framework in a stable and efficient manner. Unlike approaches that perform chunking in the actor network, we feed chunked actions into the critic network to explore potential performance gains. Our architecture leverages the Transformer's ability to process sequential information, facilitating more robust value estimation. Empirical results show that this method not only achieves efficient, stable training but also excels in sparse reward/multi-phase environments-traditionally a challenge for step-based methods. These findings underscore the promise of combining Transformer-based critics with N-returns to advance reinforcement learning performance
comment: 11 pages, 5 figures
☆ Finite-sample valid prediction of future insurance claims in the regression problem
In the current insurance literature, prediction of insurance claims in the regression problem is often performed with a statistical model. This model-based approach may suffer from several drawbacks: (i) model misspecification, (ii) selection effect, and (iii) lack of finite-sample validity. This article addresses these three issues simultaneously by employing conformal prediction-a general machine learning strategy for valid predictions. The proposed method is both model-free and tuning-parameter-free. It also guarantees finite-sample validity at a pre-assigned coverage probability level.
☆ Robust Learning of Diverse Code Edits
Software engineering activities frequently involve edits to existing code. However, contemporary code language models (LMs) lack the ability to handle diverse types of code-edit requirements. In this work, we attempt to overcome this shortcoming through (1) a novel synthetic data generation pipeline and (2) a robust model adaptation algorithm. Starting with seed code examples and diverse editing criteria, our pipeline generates high-quality samples comprising original and modified code, along with natural language instructions in different styles and verbosity. Today's code LMs come bundled with strong abilities, such as code generation and instruction following, which should not be lost due to fine-tuning. To ensure this, we propose a novel adaptation algorithm, SeleKT, that (a) leverages a dense gradient-based step to identify the weights that are most important for code editing, and (b) does a sparse projection onto the base model to avoid overfitting. Using our approach, we obtain a new series of models NextCoder (adapted from QwenCoder-2.5) that achieves strong results on five code-editing benchmarks, outperforming comparable size models and even several larger ones. We show the generality of our approach on two model families (DeepSeekCoder and QwenCoder), compare against other fine-tuning approaches, and demonstrate robustness by showing retention of code generation abilities post adaptation.
☆ Improving Neutral Point of View Text Generation through Parameter-Efficient Reinforcement Learning and a Small-Scale High-Quality Dataset
This paper describes the construction of a dataset and the evaluation of training methods to improve generative large language models' (LLMs) ability to answer queries on sensitive topics with a Neutral Point of View (NPOV), i.e., to provide significantly more informative, diverse and impartial answers. The dataset, the SHQ-NPOV dataset, comprises 300 high-quality, human-written quadruplets: a query on a sensitive topic, an answer, an NPOV rating, and a set of links to source texts elaborating the various points of view. The first key contribution of this paper is a new methodology to create such datasets through iterative rounds of human peer-critique and annotator training, which we release alongside the dataset. The second key contribution is the identification of a highly effective training regime for parameter-efficient reinforcement learning (PE-RL) to improve NPOV generation. We compare and extensively evaluate PE-RL and multiple baselines-including LoRA finetuning (a strong baseline), SFT and RLHF. PE-RL not only improves on overall NPOV quality compared to the strongest baseline ($97.06\%\rightarrow 99.08\%$), but also scores much higher on features linguists identify as key to separating good answers from the best answers ($60.25\%\rightarrow 85.21\%$ for presence of supportive details, $68.74\%\rightarrow 91.43\%$ for absence of oversimplification). A qualitative analysis corroborates this. Finally, our evaluation finds no statistical differences between results on topics that appear in the training dataset and those on separated evaluation topics, which provides strong evidence that our approach to training PE-RL exhibits very effective out of topic generalization.
☆ Limits of nonlinear and dispersive fiber propagation for photonic extreme learning
We report a generalized nonlinear Schr\"odinger equation simulation model of an extreme learning machine based on optical fiber propagation. Using handwritten digit classification as a benchmark, we study how accuracy depends on propagation dynamics, as well as parameters governing spectral encoding, readout, and noise. Test accuracies of over 91% and 93% are found for propagation in the anomalous and normal dispersion regimes respectively. Our simulation results also suggest that quantum noise on the input pulses introduces an intrinsic penalty to ELM performance.
☆ Feature Matching Intervention: Leveraging Observational Data for Causal Representation Learning
A major challenge in causal discovery from observational data is the absence of perfect interventions, making it difficult to distinguish causal features from spurious ones. We propose an innovative approach, Feature Matching Intervention (FMI), which uses a matching procedure to mimic perfect interventions. We define causal latent graphs, extending structural causal models to latent feature space, providing a framework that connects FMI with causal graph learning. Our feature matching procedure emulates perfect interventions within these causal latent graphs. Theoretical results demonstrate that FMI exhibits strong out-of-distribution (OOD) generalizability. Experiments further highlight FMI's superior performance in effectively identifying causal features solely from observational data.
☆ Deterministic Global Optimization of the Acquisition Function in Bayesian Optimization: To Do or Not To Do?
Bayesian Optimization (BO) with Gaussian Processes relies on optimizing an acquisition function to determine sampling. We investigate the advantages and disadvantages of using a deterministic global solver (MAiNGO) compared to conventional local and stochastic global solvers (L-BFGS-B and multi-start, respectively) for the optimization of the acquisition function. For CPU efficiency, we set a time limit for MAiNGO, taking the best point as optimal. We perform repeated numerical experiments, initially using the Muller-Brown potential as a benchmark function, utilizing the lower confidence bound acquisition function; we further validate our findings with three alternative benchmark functions. Statistical analysis reveals that when the acquisition function is more exploitative (as opposed to exploratory), BO with MAiNGO converges in fewer iterations than with the local solvers. However, when the dataset lacks diversity, or when the acquisition function is overly exploitative, BO with MAiNGO, compared to the local solvers, is more likely to converge to a local rather than a global ly near-optimal solution of the black-box function. L-BFGS-B and multi-start mitigate this risk in BO by introducing stochasticity in the selection of the next sampling point, which enhances the exploration of uncharted regions in the search space and reduces dependence on acquisition function hyperparameters. Ultimately, suboptimal optimization of poorly chosen acquisition functions may be preferable to their optimal solution. When the acquisition function is more exploratory, BO with MAiNGO, multi-start, and L-BFGS-B achieve comparable probabilities of convergence to a globally near-optimal solution (although BO with MAiNGO may require more iterations to converge under these conditions).
comment: 32 pages, 7 figures, 7 tables
☆ It's My Data Too: Private ML for Datasets with Multi-User Training Examples
We initiate a study of algorithms for model training with user-level differential privacy (DP), where each example may be attributed to multiple users, which we call the multi-attribution model. We first provide a carefully chosen definition of user-level DP under the multi-attribution model. Training in the multi-attribution model is facilitated by solving the contribution bounding problem, i.e. the problem of selecting a subset of the dataset for which each user is associated with a limited number of examples. We propose a greedy baseline algorithm for the contribution bounding problem. We then empirically study this algorithm for a synthetic logistic regression task and a transformer training task, including studying variants of this baseline algorithm that optimize the subset chosen using different techniques and criteria. We find that the baseline algorithm remains competitive with its variants in most settings, and build a better understanding of the practical importance of a bias-variance tradeoff inherent in solutions to the contribution bounding problem.
☆ Towards Understanding Text Hallucination of Diffusion Models via Local Generation Bias
Score-based diffusion models have achieved incredible performance in generating realistic images, audio, and video data. While these models produce high-quality samples with impressive details, they often introduce unrealistic artifacts, such as distorted fingers or hallucinated texts with no meaning. This paper focuses on textual hallucinations, where diffusion models correctly generate individual symbols but assemble them in a nonsensical manner. Through experimental probing, we consistently observe that such phenomenon is attributed it to the network's local generation bias. Denoising networks tend to produce outputs that rely heavily on highly correlated local regions, particularly when different dimensions of the data distribution are nearly pairwise independent. This behavior leads to a generation process that decomposes the global distribution into separate, independent distributions for each symbol, ultimately failing to capture the global structure, including underlying grammar. Intriguingly, this bias persists across various denoising network architectures including MLP and transformers which have the structure to model global dependency. These findings also provide insights into understanding other types of hallucinations, extending beyond text, as a result of implicit biases in the denoising models. Additionally, we theoretically analyze the training dynamics for a specific case involving a two-layer MLP learning parity points on a hypercube, offering an explanation of its underlying mechanism.
☆ PowerAttention: Exponentially Scaling of Receptive Fields for Effective Sparse Attention
Large Language Models (LLMs) face efficiency bottlenecks due to the quadratic complexity of the attention mechanism when processing long contexts. Sparse attention methods offer a promising solution, but existing approaches often suffer from incomplete effective context and/or require complex implementation of pipeline. We present a comprehensive analysis of sparse attention for autoregressive LLMs from the respective of receptive field, recognize the suboptimal nature of existing methods for expanding the receptive field, and introduce PowerAttention, a novel sparse attention design that facilitates effective and complete context extension through the theoretical analysis. PowerAttention achieves exponential receptive field growth in $d$-layer LLMs, allowing each output token to attend to $2^d$ tokens, ensuring completeness and continuity of the receptive field. Experiments demonstrate that PowerAttention outperforms existing static sparse attention methods by $5\sim 40\%$, especially on tasks demanding long-range dependencies like Passkey Retrieval and RULER, while maintaining a comparable time complexity to sliding window attention. Efficiency evaluations further highlight PowerAttention's superior speedup in both prefilling and decoding phases compared with dynamic sparse attentions and full attention ($3.0\times$ faster on 128K context), making it a highly effective and user-friendly solution for processing long sequences in LLMs.
comment: for associated code, see https://github.com/w568w/PowerAttention
☆ A Generative System for Robot-to-Human Handovers: from Intent Inference to Spatial Configuration Imagery
We propose a novel system for robot-to-human object handover that emulates human coworker interactions. Unlike most existing studies that focus primarily on grasping strategies and motion planning, our system focus on 1. inferring human handover intents, 2. imagining spatial handover configuration. The first one integrates multimodal perception-combining visual and verbal cues-to infer human intent. The second one using a diffusion-based model to generate the handover configuration, involving the spacial relationship among robot's gripper, the object, and the human hand, thereby mimicking the cognitive process of motor imagery. Experimental results demonstrate that our approach effectively interprets human cues and achieves fluent, human-like handovers, offering a promising solution for collaborative robotics. Code, videos, and data are available at: https://i3handover.github.io.
☆ Optimal Decision Tree Pruning Revisited: Algorithms and Complexity
We present a comprehensive classical and parameterized complexity analysis of decision tree pruning operations, extending recent research on the complexity of learning small decision trees. Thereby, we offer new insights into the computational challenges of decision tree simplification, a crucial aspect of developing interpretable and efficient machine learning models. We focus on fundamental pruning operations of subtree replacement and raising, which are used in heuristics. Surprisingly, while optimal pruning can be performed in polynomial time for subtree replacement, the problem is NP-complete for subtree raising. Therefore, we identify parameters and combinations thereof that lead to fixed-parameter tractability or hardness, establishing a precise borderline between these complexity classes. For example, while subtree raising is hard for small domain size $D$ or number $d$ of features, it can be solved in $D^{2d} \cdot |I|^{O(1)}$ time, where $|I|$ is the input size. We complement our theoretical findings with preliminary experimental results, demonstrating the practical implications of our analysis.
☆ Olympus: A Jumping Quadruped for Planetary Exploration Utilizing Reinforcement Learning for In-Flight Attitude Control ICRA
Exploring planetary bodies with lower gravity, such as the moon and Mars, allows legged robots to utilize jumping as an efficient form of locomotion thus giving them a valuable advantage over traditional rovers for exploration. Motivated by this fact, this paper presents the design, simulation, and learning-based "in-flight" attitude control of Olympus, a jumping legged robot tailored to the gravity of Mars. First, the design requirements are outlined followed by detailing how simulation enabled optimizing the robot's design - from its legs to the overall configuration - towards high vertical jumping, forward jumping distance, and in-flight attitude reorientation. Subsequently, the reinforcement learning policy used to track desired in-flight attitude maneuvers is presented. Successfully crossing the sim2real gap, extensive experimental studies of attitude reorientation tests are demonstrated.
comment: 7 pages, 6 figures, Accepted to the IEEE International Conference on Robotics and Automation (ICRA) 2025
☆ Domain Consistent Industrial Decarbonisation of Global Coal Power Plants
Machine learning and optimisation techniques (MLOPT) hold significant potential to accelerate the decarbonisation of industrial systems by enabling data-driven operational improvements. However, the practical application of MLOPT in industrial settings is often hindered by a lack of domain compliance and system-specific consistency, resulting in suboptimal solutions with limited real-world applicability. To address this challenge, we propose a novel human-in-the-loop (HITL) constraint-based optimisation framework that integrates domain expertise with data-driven methods, ensuring solutions are both technically sound and operationally feasible. We demonstrate the efficacy of this framework through a case study focused on enhancing the thermal efficiency and reducing the turbine heat rate of a 660 MW supercritical coal-fired power plant. By embedding domain knowledge as constraints within the optimisation process, our approach yields solutions that align with the plant's operational patterns and are seamlessly integrated into its control systems. Empirical validation confirms a mean improvement in thermal efficiency of 0.64\% and a mean reduction in turbine heat rate of 93 kJ/kWh. Scaling our analysis to 59 global coal power plants with comparable capacity and fuel type, we estimate a cumulative lifetime reduction of 156.4 million tons of carbon emissions. These results underscore the transformative potential of our HITL-MLOPT framework in delivering domain-compliant, implementable solutions for industrial decarbonisation, offering a scalable pathway to mitigate the environmental impact of coal-based power generation worldwide.
comment: 6 figures. 17 pages
☆ Probabilistic Insights for Efficient Exploration Strategies in Reinforcement Learning
We investigate efficient exploration strategies of environments with unknown stochastic dynamics and sparse rewards. Specifically, we analyze first the impact of parallel simulations on the probability of reaching rare states within a finite time budget. Using simplified models based on random walks and L\'evy processes, we provide analytical results that demonstrate a phase transition in reaching probabilities as a function of the number of parallel simulations. We identify an optimal number of parallel simulations that balances exploration diversity and time allocation. Additionally, we analyze a restarting mechanism that exponentially enhances the probability of success by redirecting efforts toward more promising regions of the state space. Our findings contribute to a more qualitative and quantitative theory of some exploration schemes in reinforcement learning, offering insights into developing more efficient strategies for environments characterized by rare events.
Transformer-Based Power Optimization for Max-Min Fairness in Cell-Free Massive MIMO
Power allocation is an important task in wireless communication networks. Classical optimization algorithms and deep learning methods, while effective in small and static scenarios, become either computationally demanding or unsuitable for large and dynamic networks with varying user loads. This letter explores the potential of transformer-based deep learning models to address these challenges. We propose a transformer neural network to jointly predict optimal uplink and downlink power using only user and access point positions. The max-min fairness problem in cell-free massive multiple input multiple output systems is considered. Numerical results show that the trained model provides near-optimal performance and adapts to varying numbers of users and access points without retraining, additional processing, or updating its neural network architecture. This demonstrates the effectiveness of the proposed model in achieving robust and flexible power allocation for dynamic networks.
comment: 5 pages, IEEE WCL, 4 FIGURES
☆ Simulation-Based Performance Evaluation of 3D Object Detection Methods with Deep Learning for a LiDAR Point Cloud Dataset in a SOTIF-related Use Case
Safety of the Intended Functionality (SOTIF) addresses sensor performance limitations and deep learning-based object detection insufficiencies to ensure the intended functionality of Automated Driving Systems (ADS). This paper presents a methodology examining the adaptability and performance evaluation of the 3D object detection methods on a LiDAR point cloud dataset generated by simulating a SOTIF-related Use Case. The major contributions of this paper include defining and modelling a SOTIF-related Use Case with 21 diverse weather conditions and generating a LiDAR point cloud dataset suitable for application of 3D object detection methods. The dataset consists of 547 frames, encompassing clear, cloudy, rainy weather conditions, corresponding to different times of the day, including noon, sunset, and night. Employing MMDetection3D and OpenPCDET toolkits, the performance of State-of-the-Art (SOTA) 3D object detection methods is evaluated and compared by testing the pre-trained Deep Learning (DL) models on the generated dataset using Average Precision (AP) and Recall metrics.
☆ Revisiting the Role of Relearning in Semantic Dementia
Patients with semantic dementia (SD) present with remarkably consistent atrophy of neurons in the anterior temporal lobe and behavioural impairments, such as graded loss of category knowledge. While relearning of lost knowledge has been shown in acute brain injuries such as stroke, it has not been widely supported in chronic cognitive diseases such as SD. Previous research has shown that deep linear artificial neural networks exhibit stages of semantic learning akin to humans. Here, we use a deep linear network to test the hypothesis that relearning during disease progression rather than particular atrophy cause the specific behavioural patterns associated with SD. After training the network to generate the common semantic features of various hierarchically organised objects, neurons are successively deleted to mimic atrophy while retraining the model. The model with relearning and deleted neurons reproduced errors specific to SD, including prototyping errors and cross-category confusions. This suggests that relearning is necessary for artificial neural networks to reproduce the behavioural patterns associated with SD in the absence of \textit{output} non-linearities. Our results support a theory of SD progression that results from continuous relearning of lost information. Future research should revisit the role of relearning as a contributing factor to cognitive diseases.
comment: 3 pages, 2 figures, presented at the Cognitive Computational Neuroscience Conference (CCN) 2023
☆ Intrinsic and Extrinsic Factor Disentanglement for Recommendation in Various Context Scenarios
In recommender systems, the patterns of user behaviors (e.g., purchase, click) may vary greatly in different contexts (e.g., time and location). This is because user behavior is jointly determined by two types of factors: intrinsic factors, which reflect consistent user preference, and extrinsic factors, which reflect external incentives that may vary in different contexts. Differentiating between intrinsic and extrinsic factors helps learn user behaviors better. However, existing studies have only considered differentiating them from a single, pre-defined context (e.g., time or location), ignoring the fact that a user's extrinsic factors may be influenced by the interplay of various contexts at the same time. In this paper, we propose the Intrinsic-Extrinsic Disentangled Recommendation (IEDR) model, a generic framework that differentiates intrinsic from extrinsic factors considering various contexts simultaneously, enabling more accurate differentiation of factors and hence the improvement of recommendation accuracy. IEDR contains a context-invariant contrastive learning component to capture intrinsic factors, and a disentanglement component to extract extrinsic factors under the interplay of various contexts. The two components work together to achieve effective factor learning. Extensive experiments on real-world datasets demonstrate IEDR's effectiveness in learning disentangled factors and significantly improving recommendation accuracy by up to 4% in NDCG.
comment: 32 pages, 13 figures, 11 tables. Accepted by Transactions of Information Systems
☆ O-RAN xApps Conflict Management using Graph Convolutional Networks
Open Radio Access Network (O-RAN) adopts a flexible, open, and virtualized structure with standardized interfaces, reducing dependency on a single supplier. Conflict management in O-RAN refers to the process of identifying and resolving conflicts between network applications. xApps are applications deployed at the RAN Intelligent Controller (RIC) that leverage advanced AI/ML algorithms to make dynamic decisions for network optimization. The lack of a unified mechanism to coordinate and prioritize the actions of different applications can create three types of conflicts (direct, indirect, and implicit). In our paper, we introduce a novel data-driven GCN-based method called Graph-based xApps Conflict and Root Cause Analysis Engine (GRACE) based on Graph Convolutional Network (GCN). It detects three types of conflicts (direct, indirect, and implicit) and pinpoints the root causes (xApps). GRACE captures the complex and hidden dependencies among the xApps, the controlled parameters, and the KPIs in O-RAN to detect possible conflicts. Then, it identifies the root causes (xApps) contributing to the detected conflicts. The proposed method was tested on highly imbalanced datasets where the number of conflict instances ranges from 40% to 10%. The model is tested in a setting that simulates real-world scenarios where conflicts are rare to assess its performance and generalizability. Experimental results demonstrate an exceptional performance, achieving a high F1-score greater than 98% for all the case studies.
comment: 9 pages, 10 figures
☆ DO-IQS: Dynamics-Aware Offline Inverse Q-Learning for Optimal Stopping with Unknown Gain Functions
We consider Inverse Optimal Stopping (IOS) problem where, based on stopped expert trajectories, one aims to recover the optimal stopping region through continuation and stopping gain functions approximation. The uniqueness of the stopping region allows the use of IOS in real-world applications with safety concerns. While current state-of-the-art inverse reinforcement learning methods recover both a Q-function and the corresponding optimal policy, they fail to account for specific challenges posed by optimal stopping problems. These include data sparsity near the stopping region, non-Markovian nature of the continuation gain, a proper treatment of boundary conditions, the need for a stable offline approach for risk-sensitive applications, and a lack of a quality evaluation metric. These challenges are addressed with the proposed Dynamics-Aware Offline Inverse Q-Learning for Optimal Stopping (DO-IQS), which incorporates temporal information by approximating the cumulative continuation gain together with the world dynamics and the Q-function without querying to the environment. Moreover, a confidence-based oversampling approach is proposed to treat the data sparsity problem. We demonstrate the performance of our models on real and artificial data including an optimal intervention for critical events problem.
☆ An Aspect Extraction Framework using Different Embedding Types, Learning Models, and Dependency Structure
Aspect-based sentiment analysis has gained significant attention in recent years due to its ability to provide fine-grained insights for sentiment expressions related to specific features of entities. An important component of aspect-based sentiment analysis is aspect extraction, which involves identifying and extracting aspect terms from text. Effective aspect extraction serves as the foundation for accurate sentiment analysis at the aspect level. In this paper, we propose aspect extraction models that use different types of embeddings for words and part-of-speech tags and that combine several learning models. We also propose tree positional encoding that is based on dependency parsing output to capture better the aspect positions in sentences. In addition, a new aspect extraction dataset is built for Turkish by machine translating an English dataset in a controlled setting. The experiments conducted on two Turkish datasets showed that the proposed models mostly outperform the studies that use the same datasets, and incorporating tree positional encoding increases the performance of the models.
comment: Aspect-based Sentiment Analysis, Aspect Extraction, Natural Language Processing, Machine Learning, Deep Neural Networks, Turkish
☆ Rethinking Synthetic Data definitions: A privacy driven approach
Synthetic data is gaining traction as a cost-effective solution for the increasing data demands of AI development and can be generated either from existing knowledge or derived data captured from real-world events. The source of the synthetic data generation and the technique used significantly impacts its residual privacy risk and therefore its opportunity for sharing. Traditional classification of synthetic data types no longer fit the newer generation techniques and there is a need to better align the classification with practical needs. We suggest a new way of grouping synthetic data types that better supports privacy evaluations to aid regulatory policymaking. Our novel classification provides flexibility to new advancements like deep generative methods and offers a more practical framework for future applications.
☆ Collaborative Expert LLMs Guided Multi-Objective Molecular Optimization
Molecular optimization is a crucial yet complex and time-intensive process that often acts as a bottleneck for drug development. Traditional methods rely heavily on trial and error, making multi-objective optimization both time-consuming and resource-intensive. Current AI-based methods have shown limited success in handling multi-objective optimization tasks, hampering their practical utilization. To address this challenge, we present MultiMol, a collaborative large language model (LLM) system designed to guide multi-objective molecular optimization. MultiMol comprises two agents, including a data-driven worker agent and a literature-guided research agent. The data-driven worker agent is a large language model being fine-tuned to learn how to generate optimized molecules considering multiple objectives, while the literature-guided research agent is responsible for searching task-related literature to find useful prior knowledge that facilitates identifying the most promising optimized candidates. In evaluations across six multi-objective optimization tasks, MultiMol significantly outperforms existing methods, achieving a 82.30% success rate, in sharp contrast to the 27.50% success rate of current strongest methods. To further validate its practical impact, we tested MultiMol on two real-world challenges. First, we enhanced the selectivity of Xanthine Amine Congener (XAC), a promiscuous ligand that binds both A1R and A2AR, successfully biasing it towards A1R. Second, we improved the bioavailability of Saquinavir, an HIV-1 protease inhibitor with known bioavailability limitations. Overall, these results indicate that MultiMol represents a highly promising approach for multi-objective molecular optimization, holding great potential to accelerate the drug development process and contribute to the advancement of pharmaceutical research.
☆ State-offset Tuning: State-based Parameter-Efficient Fine-Tuning for State Space Models
State Space Models (SSMs) have emerged as efficient alternatives to Transformers, mitigating their quadratic computational cost. However, the application of Parameter-Efficient Fine-Tuning (PEFT) methods to SSMs remains largely unexplored. In particular, prompt-based methods like Prompt Tuning and Prefix-Tuning, which are widely used in Transformers, do not perform well on SSMs. To address this, we propose state-based methods as a superior alternative to prompt-based methods. This new family of methods naturally stems from the architectural characteristics of SSMs. State-based methods adjust state-related features directly instead of depending on external prompts. Furthermore, we introduce a novel state-based PEFT method: State-offset Tuning. At every timestep, our method directly affects the state at the current step, leading to more effective adaptation. Through extensive experiments across diverse datasets, we demonstrate the effectiveness of our method. Code is available at https://github.com/furiosa-ai/ssm-state-tuning.
comment: Code is available at https://github.com/furiosa-ai/ssm-state-tuning
☆ Federated Learning for Predicting Mild Cognitive Impairment to Dementia Conversion
Dementia is a progressive condition that impairs an individual's cognitive health and daily functioning, with mild cognitive impairment (MCI) often serving as its precursor. The prediction of MCI to dementia conversion has been well studied, but previous studies have almost always focused on traditional Machine Learning (ML) based methods that require sharing sensitive clinical information to train predictive models. This study proposes a privacy-enhancing solution using Federated Learning (FL) to train predictive models for MCI to dementia conversion without sharing sensitive data, leveraging socio demographic and cognitive measures. We simulated and compared two network architectures, Peer to Peer (P2P) and client-server, to enable collaborative learning. Our results demonstrated that FL had comparable predictive performance to centralized ML, and each clinical site showed similar performance without sharing local data. Moreover, the predictive performance of FL models was superior to site specific models trained without collaboration. This work highlights that FL can eliminate the need for data sharing without compromising model efficacy.
comment: This work has been submitted to the IEEE for possible publication
☆ Differentially Private Learners for Heterogeneous Treatment Effects ICLR 2025
Patient data is widely used to estimate heterogeneous treatment effects and thus understand the effectiveness and safety of drugs. Yet, patient data includes highly sensitive information that must be kept private. In this work, we aim to estimate the conditional average treatment effect (CATE) from observational data under differential privacy. Specifically, we present DP-CATE, a novel framework for CATE estimation that is Neyman-orthogonal and further ensures differential privacy of the estimates. Our framework is highly general: it applies to any two-stage CATE meta-learner with a Neyman-orthogonal loss function, and any machine learning model can be used for nuisance estimation. We further provide an extension of our DP-CATE, where we employ RKHS regression to release the complete CATE function while ensuring differential privacy. We demonstrate our DP-CATE across various experiments using synthetic and real-world datasets. To the best of our knowledge, we are the first to provide a framework for CATE estimation that is Neyman-orthogonal and differentially private.
comment: Published at ICLR 2025
☆ TEDDY: A Family Of Foundation Models For Understanding Single Cell Biology
Understanding the biological mechanism of disease is critical for medicine, and in particular drug discovery. AI-powered analysis of genome-scale biological data hold great potential in this regard. The increasing availability of single-cell RNA sequencing data has enabled the development of large foundation models for disease biology. However, existing foundation models either do not improve or only modestly improve over task-specific models in downstream applications. Here, we explored two avenues for improving the state-of-the-art. First, we scaled the pre-training dataset to 116 million cells, which is larger than those used by previous models. Second, we leveraged the availability of large-scale biological annotations as a form of supervision during pre-training. We trained the TEDDY family of models comprising six transformer-based state-of-the-art single-cell foundation models with 70 million, 160 million, and 400 million parameters. We vetted our models on two downstream evaluation tasks -- identifying the underlying disease state of held-out donors not seen during training and distinguishing healthy cells from diseased ones for disease conditions and donors not seen during training. Scaling experiments showed that performance improved predictably with both data volume and parameter count. Our models showed substantial improvement over existing work on the first task and more muted improvements on the second.
☆ Open-Source Large Language Models as Multilingual Crowdworkers: Synthesizing Open-Domain Dialogues in Several Languages With No Examples in Targets and No Machine Translation
The prevailing paradigm in the domain of Open-Domain Dialogue agents predominantly focuses on the English language, encompassing both models and datasets. Furthermore, the financial and temporal investments required for crowdsourcing such datasets for finetuning are substantial, particularly when multiple languages are involved. Fortunately, advancements in Large Language Models (LLMs) have unveiled a plethora of possibilities across diverse tasks. Specifically, instruction-tuning has enabled LLMs to execute tasks based on natural language instructions, occasionally surpassing the performance of human crowdworkers. Additionally, these models possess the capability to function in various languages within a single thread. Consequently, to generate new samples in different languages, we propose leveraging these capabilities to replicate the data collection process. We introduce a pipeline for generating Open-Domain Dialogue data in multiple Target Languages using LLMs, with demonstrations provided in a unique Source Language. By eschewing explicit Machine Translation in this approach, we enhance the adherence to language-specific nuances. We apply this methodology to the PersonaChat dataset. To enhance the openness of generated dialogues and mimic real life scenarii, we added the notion of speech events corresponding to the type of conversation the speakers are involved in and also that of common ground which represents the premises of a conversation.
☆ Data Poisoning Attacks to Locally Differentially Private Range Query Protocols
Trajectory data, which tracks movements through geographic locations, is crucial for improving real-world applications. However, collecting such sensitive data raises considerable privacy concerns. Local differential privacy (LDP) offers a solution by allowing individuals to locally perturb their trajectory data before sharing it. Despite its privacy benefits, LDP protocols are vulnerable to data poisoning attacks, where attackers inject fake data to manipulate aggregated results. In this work, we make the first attempt to analyze vulnerabilities in several representative LDP trajectory protocols. We propose \textsc{TraP}, a heuristic algorithm for data \underline{P}oisoning attacks using a prefix-suffix method to optimize fake \underline{Tra}jectory selection, significantly reducing computational complexity. Our experimental results demonstrate that our attack can substantially increase target pattern occurrences in the perturbed trajectory dataset with few fake users. This study underscores the urgent need for robust defenses and better protocol designs to safeguard LDP trajectory data against malicious manipulation.
☆ Conceptualizing Uncertainty
Uncertainty in machine learning refers to the degree of confidence or lack thereof in a model's predictions. While uncertainty quantification methods exist, explanations of uncertainty, especially in high-dimensional settings, remain an open challenge. Existing work focuses on feature attribution approaches which are restricted to local explanations. Understanding uncertainty, its origins, and characteristics on a global scale is crucial for enhancing interpretability and trust in a model's predictions. In this work, we propose to explain the uncertainty in high-dimensional data classification settings by means of concept activation vectors which give rise to local and global explanations of uncertainty. We demonstrate the utility of the generated explanations by leveraging them to refine and improve our model.
☆ Gradient Deconfliction via Orthogonal Projections onto Subspaces For Multi-task Learning WSDM 2025
Although multi-task learning (MTL) has been a preferred approach and successfully applied in many real-world scenarios, MTL models are not guaranteed to outperform single-task models on all tasks mainly due to the negative effects of conflicting gradients among the tasks. In this paper, we fully examine the influence of conflicting gradients and further emphasize the importance and advantages of achieving non-conflicting gradients which allows simple but effective trade-off strategies among the tasks with stable performance. Based on our findings, we propose the Gradient Deconfliction via Orthogonal Projections onto Subspaces (GradOPS) spanned by other task-specific gradients. Our method not only solves all conflicts among the tasks, but can also effectively search for diverse solutions towards different trade-off preferences among the tasks. Theoretical analysis on convergence is provided, and performance of our algorithm is fully testified on multiple benchmarks in various domains. Results demonstrate that our method can effectively find multiple state-of-the-art solutions with different trade-off strategies among the tasks on multiple datasets.
comment: WSDM 2025
☆ Early-Stopped Mirror Descent for Linear Regression over Convex Bodies
Early-stopped iterative optimization methods are widely used as alternatives to explicit regularization, and direct comparisons between early-stopping and explicit regularization have been established for many optimization geometries. However, most analyses depend heavily on the specific properties of the optimization geometry or strong convexity of the empirical objective, and it remains unclear whether early-stopping could ever be less statistically efficient than explicit regularization for some particular shape constraint, especially in the overparameterized regime. To address this question, we study the setting of high-dimensional linear regression under additive Gaussian noise when the ground truth is assumed to lie in a known convex body and the task is to minimize the in-sample mean squared error. Our main result shows that for any convex body and any design matrix, up to an absolute constant factor, the worst-case risk of unconstrained early-stopped mirror descent with an appropriate potential is at most that of the least squares estimator constrained to the convex body. We achieve this by constructing algorithmic regularizers based on the Minkowski functional of the convex body.
☆ Simplicial SMOTE: Oversampling Solution to the Imbalanced Learning Problem KDD 2025
SMOTE (Synthetic Minority Oversampling Technique) is the established geometric approach to random oversampling to balance classes in the imbalanced learning problem, followed by many extensions. Its idea is to introduce synthetic data points of the minor class, with each new point being the convex combination of an existing data point and one of its k-nearest neighbors. In this paper, by viewing SMOTE as sampling from the edges of a geometric neighborhood graph and borrowing tools from the topological data analysis, we propose a novel technique, Simplicial SMOTE, that samples from the simplices of a geometric neighborhood simplicial complex. A new synthetic point is defined by the barycentric coordinates w.r.t. a simplex spanned by an arbitrary number of data points being sufficiently close rather than a pair. Such a replacement of the geometric data model results in better coverage of the underlying data distribution compared to existing geometric sampling methods and allows the generation of synthetic points of the minority class closer to the majority class on the decision boundary. We experimentally demonstrate that our Simplicial SMOTE outperforms several popular geometric sampling methods, including the original SMOTE. Moreover, we show that simplicial sampling can be easily integrated into existing SMOTE extensions. We generalize and evaluate simplicial extensions of the classic Borderline SMOTE, Safe-level SMOTE, and ADASYN algorithms, all of which outperform their graph-based counterparts.
comment: Accepted at KDD 2025 (research track)
☆ Evolutionary Prediction Games
When users decide whether to use a system based on the quality of predictions they receive, learning has the capacity to shape the population of users it serves - for better or worse. This work aims to study the long-term implications of this process through the lens of evolutionary game theory. We introduce and study evolutionary prediction games, designed to capture the role of learning as a driver of natural selection between groups of users, and hence a determinant of evolutionary outcomes. Our main theoretical results show that: (i) in settings with unlimited data and compute, learning tends to reinforce the survival of the fittest, and (ii) in more realistic settings, opportunities for coexistence emerge. We analyze these opportunities in terms of their stability and feasibility, present several mechanisms that can sustain their existence, and empirically demonstrate our findings using real and synthetic data.
comment: Comments are welcome
☆ Predicting Practically? Domain Generalization for Predictive Analytics in Real-world Environments
Predictive machine learning models are widely used in customer relationship management (CRM) to forecast customer behaviors and support decision-making. However, the dynamic nature of customer behaviors often results in significant distribution shifts between training data and serving data, leading to performance degradation in predictive models. Domain generalization, which aims to train models that can generalize to unseen environments without prior knowledge of their distributions, has become a critical area of research. In this work, we propose a novel domain generalization method tailored to handle complex distribution shifts, encompassing both covariate and concept shifts. Our method builds upon the Distributionally Robust Optimization framework, optimizing model performance over a set of hypothetical worst-case distributions rather than relying solely on the training data. Through simulation experiments, we demonstrate the working mechanism of the proposed method. We also conduct experiments on a real-world customer churn dataset, and validate its effectiveness in both temporal and spatial generalization settings. Finally, we discuss the broader implications of our method for advancing Information Systems (IS) design research, particularly in building robust predictive models for dynamic managerial environments.
☆ Multi-Agent DRL for Queue-Aware Task Offloading in Hierarchical MEC-Enabled Air-Ground Networks
Mobile edge computing (MEC)-enabled air-ground networks are a key component of 6G, employing aerial base stations (ABSs) such as unmanned aerial vehicles (UAVs) and high-altitude platform stations (HAPS) to provide dynamic services to ground IoT devices (IoTDs). These IoTDs support real-time applications (e.g., multimedia and Metaverse services) that demand high computational resources and strict quality of service (QoS) guarantees in terms of latency and task queue management. Given their limited energy and processing capabilities, IoTDs rely on UAVs and HAPS to offload tasks for distributed processing, forming a multi-tier MEC system. This paper tackles the overall energy minimization problem in MEC-enabled air-ground integrated networks (MAGIN) by jointly optimizing UAV trajectories, computing resource allocation, and queue-aware task offloading decisions. The optimization is challenging due to the nonconvex, nonlinear nature of this hierarchical system, which renders traditional methods ineffective. We reformulate the problem as a multi-agent Markov decision process (MDP) with continuous action spaces and heterogeneous agents, and propose a novel variant of multi-agent proximal policy optimization with a Beta distribution (MAPPO-BD) to solve it. Extensive simulations show that MAPPO-BD outperforms baseline schemes, achieving superior energy savings and efficient resource management in MAGIN while meeting queue delay and edge computing constraints.
☆ GNNMerge: Merging of GNN Models Without Accessing Training Data
Model merging has gained prominence in machine learning as a method to integrate multiple trained models into a single model without accessing the original training data. While existing approaches have demonstrated success in domains such as computer vision and NLP, their application to Graph Neural Networks (GNNs) remains unexplored. These methods often rely on the assumption of shared initialization, which is seldom applicable to GNNs. In this work, we undertake the first benchmarking study of model merging algorithms for GNNs, revealing their limited effectiveness in this context. To address these challenges, we propose GNNMerge, which utilizes a task-agnostic node embedding alignment strategy to merge GNNs. Furthermore, we establish that under a mild relaxation, the proposed optimization objective admits direct analytical solutions for widely used GNN architectures, significantly enhancing its computational efficiency. Empirical evaluations across diverse datasets, tasks, and architectures establish GNNMerge to be up to 24% more accurate than existing methods while delivering over 2 orders of magnitude speed-up compared to training from scratch.
☆ Paths and Ambient Spaces in Neural Loss Landscapes AISTATS 2025
Understanding the structure of neural network loss surfaces, particularly the emergence of low-loss tunnels, is critical for advancing neural network theory and practice. In this paper, we propose a novel approach to directly embed loss tunnels into the loss landscape of neural networks. Exploring the properties of these loss tunnels offers new insights into their length and structure and sheds light on some common misconceptions. We then apply our approach to Bayesian neural networks, where we improve subspace inference by identifying pitfalls and proposing a more natural prior that better guides the sampling procedure.
comment: 9 pages, Accepted at AISTATS 2025
☆ A Novel Multi-Criteria Local Latin Hypercube Refinement System for Commutation Angle Improvement in IPMSMs
The commutation angle is defined as the angle between the fundamental of the motor phase current and the fundamental of the back-EMF. It can be utilised to provide a compensating effect in IPMSMs. This is due to the reluctance torque component being dependent on the commutation angle of the phase current even before entering the extended speed range. A real-time maximum torque per current and voltage strategy is demonstrated to find the trajectory and optimum commutation angles, gamma, where the level of accuracy depends on the application and available computational speed. A magnet volume reduction using a novel multi-criteria local Latin hypercube refinement (MLHR) sampling system is also presented to improve the optimisation process. The proposed new technique minimises the magnet mass to motor torque density whilst maintaining a similar phase current level. A mapping of gamma allows the determination of the optimum angles, as shown in this paper. The 3rd generation Toyota Prius IPMSM is considered as the reference motor, where the rotor configuration is altered to allow for an individual assessment.
Transformers for molecular property prediction: Domain adaptation efficiently improves performance
Most of the current transformer-based chemical language models are pre-trained on millions to billions of molecules. However, the improvement from such scaling in dataset size is not confidently linked to improved molecular property prediction. The aim of this study is to investigate and overcome some of the limitations of transformer models in predicting molecular properties. Specifically, we examine the impact of pre-training dataset size and diversity on the performance of transformer models and investigate the use of domain adaptation as a technique for improving model performance. First, our findings indicate that increasing pretraining dataset size beyond 400K molecules from the GuacaMol dataset does not result in a significant improvement on four ADME endpoints, namely, solubility, permeability, microsomal stability, and plasma protein binding. Second, our results demonstrate that using domain adaptation by further training the transformer model on a small set of domain-relevant molecules, i.e., a few hundred to a few thousand, using multi-task regression of physicochemical properties was sufficient to significantly improve performance for three out of the four investigated ADME endpoints (P-value < 0.001). Finally, we observe that a model pre-trained on 400K molecules and domain adopted on a few hundred/thousand molecules performs similarly (P-value > 0.05) to more complicated transformer models like MolBERT(pre-trained on 1.3M molecules) and MolFormer (pre-trained on 100M molecules). A comparison to a random forest model trained on basic physicochemical properties showed similar performance to the examined transformer models. We believe that current transformer models can be improved through further systematic analysis of pre-training and downstream data, pre-training objectives, and scaling laws, ultimately leading to better and more helpful models.
☆ Video Super-Resolution: All You Need is a Video Diffusion Model
We present a generic video super-resolution algorithm in this paper, based on the Diffusion Posterior Sampling framework with an unconditional video generation model in latent space. The video generation model, a diffusion transformer, functions as a space-time model. We argue that a powerful model, which learns the physics of the real world, can easily handle various kinds of motion patterns as prior knowledge, thus eliminating the need for explicit estimation of optical flows or motion parameters for pixel alignment. Furthermore, a single instance of the proposed video diffusion transformer model can adapt to different sampling conditions without re-training. Due to limited computational resources and training data, our experiments provide empirical evidence of the algorithm's strong super-resolution capabilities using synthetic data.
☆ Leap: Inductive Link Prediction via Learnable TopologyAugmentation
Link prediction is a crucial task in many downstream applications of graph machine learning. To this end, Graph Neural Network (GNN) is a widely used technique for link prediction, mainly in transductive settings, where the goal is to predict missing links between existing nodes. However, many real-life applications require an inductive setting that accommodates for new nodes, coming into an existing graph. Thus, recently inductive link prediction has attracted considerable attention, and a multi-layer perceptron (MLP) is the popular choice of most studies to learn node representations. However, these approaches have limited expressivity and do not fully capture the graph's structural signal. Therefore, in this work we propose LEAP, an inductive link prediction method based on LEArnable toPology augmentation. Unlike previous methods, LEAP models the inductive bias from both the structure and node features, and hence is more expressive. To the best of our knowledge, this is the first attempt to provide structural contexts for new nodes via learnable augmentation in inductive settings. Extensive experiments on seven real-world homogeneous and heterogeneous graphs demonstrates that LEAP significantly surpasses SOTA methods. The improvements are up to 22\% and 17\% in terms of AUC and average precision, respectively. The code and datasets are available on GitHub (https://github.com/AhmedESamy/LEAP/)
comment: published in Machine Learning, Optimization, and Data Science, Springer Nature Switzerland
LLM as GNN: Graph Vocabulary Learning for Text-Attributed Graph Foundation Models
Text-Attributed Graphs (TAGs), where each node is associated with text descriptions, are ubiquitous in real-world scenarios. They typically exhibit distinctive structure and domain-specific knowledge, motivating the development of a Graph Foundation Model (GFM) that generalizes across diverse graphs and tasks. Despite large efforts to integrate Large Language Models (LLMs) and Graph Neural Networks (GNNs) for TAGs, existing approaches suffer from decoupled architectures with two-stage alignment, limiting their synergistic potential. Even worse, existing methods assign out-of-vocabulary (OOV) tokens to graph nodes, leading to graph-specific semantics, token explosion, and incompatibility with task-oriented prompt templates, which hinders cross-graph and cross-task transferability. To address these challenges, we propose PromptGFM, a versatile GFM for TAGs grounded in graph vocabulary learning. PromptGFM comprises two key components: (1) Graph Understanding Module, which explicitly prompts LLMs to replicate the finest GNN workflow within the text space, facilitating seamless GNN-LLM integration and elegant graph-text alignment; (2) Graph Inference Module, which establishes a language-based graph vocabulary ensuring expressiveness, transferability, and scalability, enabling readable instructions for LLM fine-tuning. Extensive experiments demonstrate our superiority and transferability across diverse graphs and tasks. The code is available at this: https://github.com/agiresearch/PromptGFM.
☆ Differential Machine Learning for Time Series Prediction
Accurate time series prediction is challenging due to the inherent nonlinearity and sensitivity to initial conditions. We propose a novel approach that enhances neural network predictions through differential learning, which involves training models on both the original time series and its differential series. Specifically, we develop a differential long short-term memory (Diff-LSTM) network that uses a shared LSTM cell to simultaneously process both data streams, effectively capturing intrinsic patterns and temporal dynamics. Evaluated on the Mackey-Glass, Lorenz, and R\"ossler chaotic time series, as well as a real-world financial dataset from ACI Worldwide Inc., our results demonstrate that the Diff- LSTM network outperforms prevalent models such as recurrent neural networks, convolutional neural networks, and bidirectional and encoder-decoder LSTM networks in both short-term and long-term predictions. This framework offers a promising solution for enhancing time series prediction, even when comprehensive knowledge of the underlying dynamics of the time series is not fully available.
☆ Enhancing Vietnamese VQA through Curriculum Learning on Raw and Augmented Text Representations AAAI-25
Visual Question Answering (VQA) is a multimodal task requiring reasoning across textual and visual inputs, which becomes particularly challenging in low-resource languages like Vietnamese due to linguistic variability and the lack of high-quality datasets. Traditional methods often rely heavily on extensive annotated datasets, computationally expensive pipelines, and large pre-trained models, specifically in the domain of Vietnamese VQA, limiting their applicability in such scenarios. To address these limitations, we propose a training framework that combines a paraphrase-based feature augmentation module with a dynamic curriculum learning strategy. Explicitly, augmented samples are considered "easy" while raw samples are regarded as "hard". The framework then utilizes a mechanism that dynamically adjusts the ratio of easy to hard samples during training, progressively modifying the same dataset to increase its difficulty level. By enabling gradual adaptation to task complexity, this approach helps the Vietnamese VQA model generalize well, thus improving overall performance. Experimental results show consistent improvements on the OpenViVQA dataset and mixed outcomes on the ViVQA dataset, highlighting both the potential and challenges of our approach in advancing VQA for Vietnamese language.
comment: 10 pages, 3 figures, AAAI-25 Workshop on Document Understanding and Intelligence
☆ Exploring specialization and sensitivity of convolutional neural networks in the context of simultaneous image augmentations
Drawing parallels with the way biological networks are studied, we adapt the treatment--control paradigm to explainable artificial intelligence research and enrich it through multi-parametric input alterations. In this study, we propose a framework for investigating the internal inference impacted by input data augmentations. The internal changes in network operation are reflected in activation changes measured by variance, which can be decomposed into components related to each augmentation, employing Sobol indices and Shapley values. These quantities enable one to visualize sensitivity to different variables and use them for guided masking of activations. In addition, we introduce a way of single-class sensitivity analysis where the candidates are filtered according to their matching to prediction bias generated by targeted damaging of the activations. Relying on the observed parallels, we assume that the developed framework can potentially be transferred to studying biological neural networks in complex environments.
comment: 26 pages; main text: 5 figures, 4 tables; appendix: 4 sections, 3 tables; supplementary: 7 files (figures S1-S6: packed as 7z archive, S7: single pdf file)
☆ TrafficKAN-GCN: Graph Convolutional-based Kolmogorov-Arnold Network for Traffic Flow Optimization
Urban traffic optimization is critical for improving transportation efficiency and alleviating congestion, particularly in large-scale dynamic networks. Traditional methods, such as Dijkstra's and Floyd's algorithms, provide effective solutions in static settings, but they struggle with the spatial-temporal complexity of real-world traffic flows. In this work, we propose TrafficKAN-GCN, a hybrid deep learning framework combining Kolmogorov-Arnold Networks (KAN) with Graph Convolutional Networks (GCN), designed to enhance urban traffic flow optimization. By integrating KAN's adaptive nonlinear function approximation with GCN's spatial graph learning capabilities, TrafficKAN-GCN captures both complex traffic patterns and topological dependencies. We evaluate the proposed framework using real-world traffic data from the Baltimore Metropolitan area. Compared with baseline models such as MLP-GCN, standard GCN, and Transformer-based approaches, TrafficKAN-GCN achieves competitive prediction accuracy while demonstrating improved robustness in handling noisy and irregular traffic data. Our experiments further highlight the framework's ability to redistribute traffic flow, mitigate congestion, and adapt to disruptive events, such as the Francis Scott Key Bridge collapse. This study contributes to the growing body of work on hybrid graph learning for intelligent transportation systems, highlighting the potential of combining KAN and GCN for real-time traffic optimization. Future work will focus on reducing computational overhead and integrating Transformer-based temporal modeling for enhanced long-term traffic prediction. The proposed TrafficKAN-GCN framework offers a promising direction for data-driven urban mobility management, balancing predictive accuracy, robustness, and computational efficiency.
comment: 21 pages, 14 figures
☆ Benchmarking Dynamic SLO Compliance in Distributed Computing Continuum Systems
Ensuring Service Level Objectives (SLOs) in large-scale architectures, such as Distributed Computing Continuum Systems (DCCS), is challenging due to their heterogeneous nature and varying service requirements across different devices and applications. Additionally, unpredictable workloads and resource limitations lead to fluctuating performance and violated SLOs. To improve SLO compliance in DCCS, one possibility is to apply machine learning; however, the design choices are often left to the developer. To that extent, we provide a benchmark of Active Inference -- an emerging method from neuroscience -- against three established reinforcement learning algorithms (Deep Q-Network, Advantage Actor-Critic, and Proximal Policy Optimization). We consider a realistic DCCS use case: an edge device running a video conferencing application alongside a WebSocket server streaming videos. Using one of the respective algorithms, we continuously monitor key performance metrics, such as latency and bandwidth usage, to dynamically adjust parameters -- including the number of streams, frame rate, and resolution -- to optimize service quality and user experience. To test algorithms' adaptability to constant system changes, we simulate dynamically changing SLOs and both instant and gradual data-shift scenarios, such as network bandwidth limitations and fluctuating device thermal states. Although the evaluated algorithms all showed advantages and limitations, our findings demonstrate that Active Inference is a promising approach for ensuring SLO compliance in DCCS, offering lower memory usage, stable CPU utilization, and fast convergence.
☆ Conformal Transformations for Symmetric Power Transformers SC
Transformers with linear attention offer significant computational advantages over softmax-based transformers but often suffer from degraded performance. The symmetric power (sympow) transformer, a particular type of linear transformer, addresses some of this performance gap by leveraging symmetric tensor embeddings, achieving comparable performance to softmax transformers. However, the finite capacity of the recurrent state in sympow transformers limits their ability to retain information, leading to performance degradation when scaling the training or evaluation context length. To address this issue, we propose the conformal-sympow transformer, which dynamically frees up capacity using data-dependent multiplicative gating and adaptively stores information using data-dependent rotary embeddings. Preliminary experiments on the LongCrawl64 dataset demonstrate that conformal-sympow overcomes the limitations of sympow transformers, achieving robust performance across scaled training and evaluation contexts.
comment: SCOPE Workshop at ICLR 2025
☆ Trajectory Prediction for Autonomous Driving: Progress, Limitations, and Future Directions
As the potential for autonomous vehicles to be integrated on a large scale into modern traffic systems continues to grow, ensuring safe navigation in dynamic environments is crucial for smooth integration. To guarantee safety and prevent collisions, autonomous vehicles must be capable of accurately predicting the trajectories of surrounding traffic agents. Over the past decade, significant efforts from both academia and industry have been dedicated to designing solutions for precise trajectory forecasting. These efforts have produced a diverse range of approaches, raising questions about the differences between these methods and whether trajectory prediction challenges have been fully addressed. This paper reviews a substantial portion of recent trajectory prediction methods and devises a taxonomy to classify existing solutions. A general overview of the prediction pipeline is also provided, covering input and output modalities, modeling features, and prediction paradigms discussed in the literature. In addition, the paper discusses active research areas within trajectory prediction, addresses the posed research questions, and highlights the remaining research gaps and challenges.
☆ Exploring the Potential of Large Language Models as Predictors in Dynamic Text-Attributed Graphs
With the rise of large language models (LLMs), there has been growing interest in Graph Foundation Models (GFMs) for graph-based tasks. By leveraging LLMs as predictors, GFMs have demonstrated impressive generalizability across various tasks and datasets. However, existing research on LLMs as predictors has predominantly focused on static graphs, leaving their potential in dynamic graph prediction unexplored. In this work, we pioneer using LLMs for predictive tasks on dynamic graphs. We identify two key challenges: the constraints imposed by context length when processing large-scale historical data and the significant variability in domain characteristics, both of which complicate the development of a unified predictor. To address these challenges, we propose the GraphAgent-Dynamic (GAD) Framework, a multi-agent system that leverages collaborative LLMs. In contrast to using a single LLM as the predictor, GAD incorporates global and local summary agents to generate domain-specific knowledge, enhancing its transferability across domains. Additionally, knowledge reflection agents enable adaptive updates to GAD's knowledge, maintaining a unified and self-consistent architecture. In experiments, GAD demonstrates performance comparable to or even exceeds that of full-supervised graph neural networks without dataset-specific training. Finally, to enhance the task-specific performance of LLM-based predictors, we discuss potential improvements, such as dataset-specific fine-tuning to LLMs. By developing tailored strategies for different tasks, we provide new insights for the future design of LLM-based predictors.
☆ Less is more? Rewards in RL for Cyber Defence
The last few years has seen an explosion of interest in autonomous cyber defence agents based on deep reinforcement learning. Such agents are typically trained in a cyber gym environment, also known as a cyber simulator, at least 32 of which have already been built. Most, if not all cyber gyms provide dense "scaffolded" reward functions which combine many penalties or incentives for a range of (un)desirable states and costly actions. Whilst dense rewards help alleviate the challenge of exploring complex environments, yielding seemingly effective strategies from relatively few environment steps; they are also known to bias the solutions an agent can find, potentially towards suboptimal solutions. Sparse rewards could offer preferable or more effective solutions and have been overlooked by cyber gyms to date. In this work we set out to evaluate whether sparse reward functions might enable training more effective cyber defence agents. Towards this goal we first break down several evaluation limitations in existing work by proposing a ground truth evaluation score that goes beyond the standard RL paradigm used to train and evaluate agents. By adapting a well-established cyber gym to accommodate our methodology and ground truth score, we propose and evaluate two sparse reward mechanisms and compare them with a typical dense reward. Our evaluation considers a range of network sizes, from 2 to 50 nodes, and both reactive and proactive defensive actions. Our results show that sparse rewards, particularly positive reinforcement for an uncompromised network state, enable the training of more effective cyber defence agents. Furthermore, we show that sparse rewards provide more stable training than dense rewards, and that both effectiveness and training stability are robust to a variety of cyber environment considerations.
comment: 4 Pages
☆ Structural Entropy Guided Unsupervised Graph Out-Of-Distribution Detection AAAI 2025
With the emerging of huge amount of unlabeled data, unsupervised out-of-distribution (OOD) detection is vital for ensuring the reliability of graph neural networks (GNNs) by identifying OOD samples from in-distribution (ID) ones during testing, where encountering novel or unknown data is inevitable. Existing methods often suffer from compromised performance due to redundant information in graph structures, which impairs their ability to effectively differentiate between ID and OOD data. To address this challenge, we propose SEGO, an unsupervised framework that integrates structural entropy into OOD detection regarding graph classification. Specifically, within the architecture of contrastive learning, SEGO introduces an anchor view in the form of coding tree by minimizing structural entropy. The obtained coding tree effectively removes redundant information from graphs while preserving essential structural information, enabling the capture of distinct graph patterns between ID and OOD samples. Furthermore, we present a multi-grained contrastive learning scheme at local, global, and tree levels using triplet views, where coding trees with essential information serve as the anchor view. Extensive experiments on real-world datasets validate the effectiveness of SEGO, demonstrating superior performance over state-of-the-art baselines in OOD detection. Specifically, our method achieves the best performance on 9 out of 10 dataset pairs, with an average improvement of 3.7\% on OOD detection datasets, significantly surpassing the best competitor by 10.8\% on the FreeSolv/ToxCast dataset pair.
comment: Accepted by AAAI 2025 (The 39th Annual AAAI Conference on Artificial Intelligence)
☆ PAIR: A Novel Large Language Model-Guided Selection Strategy for Evolutionary Algorithms
Evolutionary Algorithms (EAs) employ random or simplistic selection methods, limiting their exploration of solution spaces and convergence to optimal solutions. The randomness in performing crossover or mutations may limit the model's ability to evolve efficiently. This paper introduces Preference-Aligned Individual Reciprocity (PAIR), a novel selection approach leveraging Large Language Models to emulate human-like mate selection, thereby introducing intelligence to the pairing process in EAs. PAIR prompts an LLM to evaluate individuals within a population based on genetic diversity, fitness level, and crossover compatibility, guiding more informed pairing decisions. We evaluated PAIR against a baseline method called LLM-driven EA (LMEA), published recently. Results indicate that PAIR significantly outperforms LMEA across various TSP instances, achieving lower optimality gaps and improved convergence. This performance is especially noticeable when combined with the flash thinking model, demonstrating increased population diversity to escape local optima. In general, PAIR provides a new strategy in the area of in-context learning for LLM-driven selection in EAs via sophisticated preference modelling, paving the way for improved solutions and further studies into LLM-guided optimization.
☆ Prediction of Halo Coronal Mass Ejections Using SDO/HMI Vector Magnetic Data Products and a Transformer Model
We present a transformer model, named DeepHalo, to predict the occurrence of halo coronal mass ejections (CMEs). Our model takes as input an active region (AR) and a profile, where the profile contains a time series of data samples in the AR that are collected 24 hours before the beginning of a day, and predicts whether the AR would produce a halo CME during that day. Each data sample contains physical parameters, or features, derived from photospheric vector magnetic field data taken by the Helioseismic and Magnetic Imager (HMI) on board the Solar Dynamics Observatory (SDO). We survey and match CME events in the Space Weather Database Of Notification, Knowledge, Information (DONKI) and Large Angle and Spectrometric Coronagraph (LASCO) CME Catalog, and compile a list of CMEs including halo CMEs and non-halo CMEs associated with ARs in the period between November 2010 and August 2023. We use the information gathered above to build the labels (positive versus negative) of the data samples and profiles at hand, where the labels are needed for machine learning. Experimental results show that DeepHalo with a true skill statistics (TSS) score of 0.907 outperforms a closely related long short-term memory network with a TSS score of 0.821. To our knowledge, this is the first time that the transformer model has been used for halo CME prediction.
comment: 13 pages, 8 figures
☆ Convergence Rates for Softmax Gating Mixture of Experts ICML 2024
Mixture of experts (MoE) has recently emerged as an effective framework to advance the efficiency and scalability of machine learning models by softly dividing complex tasks among multiple specialized sub-models termed experts. Central to the success of MoE is an adaptive softmax gating mechanism which takes responsibility for determining the relevance of each expert to a given input and then dynamically assigning experts their respective weights. Despite its widespread use in practice, a comprehensive study on the effects of the softmax gating on the MoE has been lacking in the literature. To bridge this gap in this paper, we perform a convergence analysis of parameter estimation and expert estimation under the MoE equipped with the standard softmax gating or its variants, including a dense-to-sparse gating and a hierarchical softmax gating, respectively. Furthermore, our theories also provide useful insights into the design of sample-efficient expert structures. In particular, we demonstrate that it requires polynomially many data points to estimate experts satisfying our proposed \emph{strong identifiability} condition, namely a commonly used two-layer feed-forward network. In stark contrast, estimating linear experts, which violate the strong identifiability condition, necessitates exponentially many data points as a result of intrinsic parameter interactions expressed in the language of partial differential equations. All the theoretical results are substantiated with a rigorous guarantee.
comment: Section 2 of this work comes from our previous paper titled "On Least Square Estimation in Softmax Gating Mixture of Experts" and published at the ICML 2024
☆ NodeReg: Mitigating the Imbalance and Distribution Shift Effects in Semi-Supervised Node Classification via Norm Consistency
Aggregating information from neighboring nodes benefits graph neural networks (GNNs) in semi-supervised node classification tasks. Nevertheless, this mechanism also renders nodes susceptible to the influence of their neighbors. For instance, this will occur when the neighboring nodes are imbalanced or the neighboring nodes contain noise, which can even affect the GNN's ability to generalize out of distribution. We find that ensuring the consistency of the norm for node representations can significantly reduce the impact of these two issues on GNNs. To this end, we propose a regularized optimization method called NodeReg that enforces the consistency of node representation norms. This method is simple but effective and satisfies Lipschitz continuity, thus facilitating stable optimization and significantly improving semi-supervised node classification performance under the above two scenarios. To illustrate, in the imbalance scenario, when training a GCN with an imbalance ratio of 0.1, NodeReg outperforms the most competitive baselines by 1.4%-25.9% in F1 score across five public datasets. Similarly, in the distribution shift scenario, NodeReg outperforms the most competitive baseline by 1.4%-3.1% in accuracy.
☆ An Analytical Theory of Power Law Spectral Bias in the Learning Dynamics of Diffusion Models
We developed an analytical framework for understanding how the learned distribution evolves during diffusion model training. Leveraging the Gaussian equivalence principle, we derived exact solutions for the gradient-flow dynamics of weights in one- or two-layer linear denoiser settings with arbitrary data. Remarkably, these solutions allowed us to derive the generated distribution in closed form and its KL divergence through training. These analytical results expose a pronounced power-law spectral bias, i.e., for weights and distributions, the convergence time of a mode follows an inverse power law of its variance. Empirical experiments on both Gaussian and image datasets demonstrate that the power-law spectral bias remains robust even when using deeper or convolutional architectures. Our results underscore the importance of the data covariance in dictating the order and rate at which diffusion models learn different modes of the data, providing potential explanations for why earlier stopping could lead to incorrect details in image generative models.
comment: 50 pages, 10 figures. Preprint
☆ Directly Follows Graphs Go Predictive Process Monitoring With Graph Neural Networks
In the past years, predictive process monitoring (PPM) techniques based on artificial neural networks have evolved as a method to monitor the future behavior of business processes. Existing approaches mostly focus on interpreting the processes as sequences, so-called traces, and feeding them to neural architectures designed to operate on sequential data such as recurrent neural networks (RNNs) or transformers. In this study, we investigate an alternative way to perform PPM: by transforming each process in its directly-follows-graph (DFG) representation we are able to apply graph neural networks (GNNs) for the prediction tasks. By this, we aim to develop models that are more suitable for complex processes that are long and contain an abundance of loops. In particular, we present different ways to create DFG representations depending on the particular GNN we use. The tested GNNs range from classical node-based to novel edge-based architectures. Further, we investigate the possibility of using multi-graphs. By these steps, we aim to design graph representations that minimize the information loss when transforming traces into graphs.
comment: 10 pages, 4 figures, 3 tables
☆ Online Bidding under RoS Constraints without Knowing the Value
We consider the problem of bidding in online advertising, where an advertiser aims to maximize value while adhering to budget and Return-on-Spend (RoS) constraints. Unlike prior work that assumes knowledge of the value generated by winning each impression ({e.g.,} conversions), we address the more realistic setting where the advertiser must simultaneously learn the optimal bidding strategy and the value of each impression opportunity. This introduces a challenging exploration-exploitation dilemma: the advertiser must balance exploring different bids to estimate impression values with exploiting current knowledge to bid effectively. To address this, we propose a novel Upper Confidence Bound (UCB)-style algorithm that carefully manages this trade-off. Via a rigorous theoretical analysis, we prove that our algorithm achieves $\widetilde{O}(\sqrt{T\log(|\mathcal{B}|T)})$ regret and constraint violation, where $T$ is the number of bidding rounds and $\mathcal{B}$ is the domain of possible bids. This establishes the first optimal regret and constraint violation bounds for bidding in the online setting with unknown impression values. Moreover, our algorithm is computationally efficient and simple to implement. We validate our theoretical findings through experiments on synthetic data, demonstrating that our algorithm exhibits strong empirical performance compared to existing approaches.
♻ ☆ Multiaccuracy and Multicalibration via Proxy Groups
As the use of predictive machine learning algorithms increases in high-stakes decision-making, it is imperative that these algorithms are fair across sensitive groups. Unfortunately, measuring and enforcing fairness in real-world applications can be challenging due to missing or incomplete sensitive group data. Proxy-sensitive attributes have been proposed as a practical and effective solution in these settings, but only for parity-based fairness notions. Knowing how to evaluate and control for fairness with missing sensitive group data for newer and more flexible frameworks, such as multiaccuracy and multicalibration, remains unexplored. In this work, we address this gap by demonstrating that in the absence of sensitive group data, proxy-sensitive attributes can provably be used to derive actionable upper bounds on the true multiaccuracy and multicalibration, providing insights into a model's potential worst-case fairness violations. Additionally, we show that adjusting models to satisfy multiaccuracy and multicalibration across proxy-sensitive attributes can significantly mitigate these violations for the true, but unknown, sensitive groups. Through several experiments on real-world datasets, we illustrate that approximate multiaccuracy and multicalibration can be achieved even when sensitive group information is incomplete or unavailable.
♻ ☆ ArcPro: Architectural Programs for Structured 3D Abstraction of Sparse Points CVPR 2025
We introduce ArcPro, a novel learning framework built on architectural programs to recover structured 3D abstractions from highly sparse and low-quality point clouds. Specifically, we design a domain-specific language (DSL) to hierarchically represent building structures as a program, which can be efficiently converted into a mesh. We bridge feedforward and inverse procedural modeling by using a feedforward process for training data synthesis, allowing the network to make reverse predictions. We train an encoder-decoder on the points-program pairs to establish a mapping from unstructured point clouds to architectural programs, where a 3D convolutional encoder extracts point cloud features and a transformer decoder autoregressively predicts the programs in a tokenized form. Inference by our method is highly efficient and produces plausible and faithful 3D abstractions. Comprehensive experiments demonstrate that ArcPro outperforms both traditional architectural proxy reconstruction and learning-based abstraction methods. We further explore its potential to work with multi-view image and natural language inputs.
comment: CVPR 2025 (Patent Protected); Project page: https://vcc.tech/research/2025/ArcPro
♻ ☆ Revisiting Random Walks for Learning on Graphs
We revisit a simple model class for machine learning on graphs, where a random walk on a graph produces a machine-readable record, and this record is processed by a deep neural network to directly make vertex-level or graph-level predictions. We call these stochastic machines random walk neural networks (RWNNs), and through principled analysis, show that we can design them to be isomorphism invariant while capable of universal approximation of graph functions in probability. A useful finding is that almost any kind of record of random walks guarantees probabilistic invariance as long as the vertices are anonymized. This enables us, for example, to record random walks in plain text and adopt a language model to read these text records to solve graph tasks. We further establish a parallelism to message passing neural networks using tools from Markov chain theory, and show that over-smoothing in message passing is alleviated by construction in RWNNs, while over-squashing manifests as probabilistic under-reaching. We empirically demonstrate RWNNs on a range of problems, verifying our theoretical analysis and demonstrating the use of language models for separating strongly regular graphs where 3-WL test fails, and transductive classification on arXiv citation network. Code is available at https://github.com/jw9730/random-walk.
comment: 51 pages, 14 figures
♻ ☆ BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modelling
Time-series Generation (TSG) is a prominent research area with broad applications in simulations, data augmentation, and counterfactual analysis. While existing methods have shown promise in unconditional single-domain TSG, real-world applications demand for cross-domain approaches capable of controlled generation tailored to domain-specific constraints and instance-level requirements. In this paper, we argue that text can provide semantic insights, domain information and instance-specific temporal patterns, to guide and improve TSG. We introduce ``Text-Controlled TSG'', a task focused on generating realistic time series by incorporating textual descriptions. To address data scarcity in this setting, we propose a novel LLM-based Multi-Agent framework that synthesizes diverse, realistic text-to-TS datasets. Furthermore, we introduce BRIDGE, a hybrid text-controlled TSG framework that integrates semantic prototypes with text description for supporting domain-level guidance. This approach achieves state-of-the-art generation fidelity on 11 of 12 datasets, and improves controllability by 12.52% on MSE and 6.34% MAE compared to no text input generation, highlighting its potential for generating tailored time-series data.
comment: Preprint. Work in progress
♻ ☆ SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction ICLR 2025
Large Language Models (LLMs) have demonstrated improved generation performance by incorporating externally retrieved knowledge, a process known as retrieval-augmented generation (RAG). Despite the potential of this approach, existing studies evaluate RAG effectiveness by 1) assessing retrieval and generation components jointly, which obscures retrieval's distinct contribution, or 2) examining retrievers using traditional metrics such as NDCG, which creates a gap in understanding retrieval's true utility in the overall generation process. To address the above limitations, in this work, we introduce an automatic evaluation method that measures retrieval quality through the lens of information gain within the RAG framework. Specifically, we propose Semantic Perplexity (SePer), a metric that captures the LLM's internal belief about the correctness of the retrieved information. We quantify the utility of retrieval by the extent to which it reduces semantic perplexity post-retrieval. Extensive experiments demonstrate that SePer not only aligns closely with human preferences but also offers a more precise and efficient evaluation of retrieval utility across diverse RAG scenarios.
comment: ICLR 2025 Spotlight
♻ ☆ Iterative Value Function Optimization for Guided Decoding
While Reinforcement Learning from Human Feedback (RLHF) has become the predominant method for controlling language model outputs, it suffers from high computational costs and training instability. Guided decoding, especially value-guided methods, offers a cost-effective alternative by controlling outputs without re-training models. However, the accuracy of the value function is crucial for value-guided decoding, as inaccuracies can lead to suboptimal decision-making and degraded performance. Existing methods struggle with accurately estimating the optimal value function, leading to less effective control. We propose Iterative Value Function Optimization, a novel framework that addresses these limitations through two key components: Monte Carlo Value Estimation, which reduces estimation variance by exploring diverse trajectories, and Iterative On-Policy Optimization, which progressively improves value estimation through collecting trajectories from value-guided policies. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of value-guided decoding approaches in aligning language models. These approaches not only achieve alignment but also significantly reduce computational costs by leveraging principled value function optimization for efficient and effective control.
comment: 20 pages, 10 figures
♻ ☆ Personalize Your LLM: Fake it then Align it NAACL 2025
Personalizing large language models (LLMs) is essential for delivering tailored interactions that improve user experience. Many existing personalization methods require fine-tuning LLMs for each user, rendering them prohibitively expensive for widespread adoption. Although retrieval-based approaches offer a more compute-efficient alternative, they still depend on large, high-quality datasets that are not consistently available for all users. To address this challenge, we propose CHAMELEON, a scalable and efficient personalization approach that uses (1) self-generated personal preference data and (2) representation editing to enable quick and cost-effective personalization. Our experiments on various tasks, including those from the LaMP personalization benchmark, show that CHAMELEON efficiently adapts models to personal preferences, improving instruction-tuned models and outperforms two personalization baselines by an average of 40% across two model architectures.
comment: NAACL 2025 Findings
♻ ☆ On Discriminative Probabilistic Modeling for Self-Supervised Representation Learning ICLR 2025
We study the discriminative probabilistic modeling on a continuous domain for the data prediction task of (multimodal) self-supervised representation learning. To address the challenge of computing the integral in the partition function for each anchor data, we leverage the multiple importance sampling (MIS) technique for robust Monte Carlo integration, which can recover InfoNCE-based contrastive loss as a special case. Within this probabilistic modeling framework, we conduct generalization error analysis to reveal the limitation of current InfoNCE-based contrastive loss for self-supervised representation learning and derive insights for developing better approaches by reducing the error of Monte Carlo integration. To this end, we propose a novel non-parametric method for approximating the sum of conditional probability densities required by MIS through convex optimization, yielding a new contrastive objective for self-supervised representation learning. Moreover, we design an efficient algorithm for solving the proposed objective. We empirically compare our algorithm to representative baselines on the contrastive image-language pretraining task. Experimental results on the CC3M and CC12M datasets demonstrate the superior overall performance of our algorithm. Our code is available at https://github.com/bokun-wang/NUCLR.
comment: To appear in ICLR 2025
♻ ☆ PARAMANU-GANITA: Can Small Math Language Models Rival with Large Language Models on Mathematical Reasoning?
In this paper, we study whether domain specific pretraining of small generative language models (SLM) from scratch with domain specialized tokenizer and Chain-of-Thought (CoT) instruction fine-tuning results in competitive performance on mathematical reasoning compared to LLMs? Secondly, whether this approach is environmentally sustainable, highly cost efficient? To address these research questions, we present Paramanu-Ganita, a 208 million-parameter novel decoder-only Auto Regressive SLM on mathematics. We performed pretraining from scratch on 31.5 billion tokens for 170 A100 hours using a context size of 4096 on a mixed mathematical corpus consisting of web pages, source code, textbooks, CoT templatised StackOverflow QA pairs, and mathematical lecture notes in LaTeX curated by us. We also trained a math and code specialised BPE tokenizer. We proposed and performed CoT instruction fine-tuning of Paramanu-Ganita on the MetaMathQA dataset. Our model Paramanu-Ganita, despite being 34 times smaller than the 7B LLMs, outperforms generalist LLMs by approximately 30% points, and even math-specialised LLMs by 3-23% points in GSM8K test accuracy metric. On MATH benchmark, Paramanu-Ganita outperformed the various models by 6-8% points. On benchmarks like LogiQA, MMLU (high school, college level), and competitive exams level, AGIEVAL (AQuA-RAT, SAT-Math), Paramanu-Ganita outperformed others by 1-4%. Our model is available at https://huggingface.co/gyanai/paramanu-ganita-208M-hf .
♻ ☆ Reusing Historical Trajectories in Natural Policy Gradient via Importance Sampling: Convergence and Convergence Rate
Reinforcement learning provides a mathematical framework for learning-based control, whose success largely depends on the amount of data it can utilize. The efficient utilization of historical trajectories obtained from previous policies is essential for expediting policy optimization. Empirical evidence has shown that policy gradient methods based on importance sampling work well. However, existing literature often neglect the interdependence between trajectories from different iterations, and the good empirical performance lacks a rigorous theoretical justification. In this paper, we study a variant of the natural policy gradient method with reusing historical trajectories via importance sampling. We show that the bias of the proposed estimator of the gradient is asymptotically negligible, the resultant algorithm is convergent, and reusing past trajectories helps improve the convergence rate. We further apply the proposed estimator to popular policy optimization algorithms such as trust region policy optimization. Our theoretical results are verified on classical benchmarks.
♻ ☆ Neural DNF-MT: A Neuro-symbolic Approach for Learning Interpretable and Editable Policies AAMAS 2025
Although deep reinforcement learning has been shown to be effective, the model's black-box nature presents barriers to direct policy interpretation. To address this problem, we propose a neuro-symbolic approach called neural DNF-MT for end-to-end policy learning. The differentiable nature of the neural DNF-MT model enables the use of deep actor-critic algorithms for training. At the same time, its architecture is designed so that trained models can be directly translated into interpretable policies expressed as standard (bivalent or probabilistic) logic programs. Moreover, additional layers can be included to extract abstract features from complex observations, acting as a form of predicate invention. The logic representations are highly interpretable, and we show how the bivalent representations of deterministic policies can be edited and incorporated back into a neural model, facilitating manual intervention and adaptation of learned policies. We evaluate our approach on a range of tasks requiring learning deterministic or stochastic behaviours from various forms of observations. Our empirical results show that our neural DNF-MT model performs at the level of competing black-box methods whilst providing interpretable policies.
comment: AAMAS 2025 (with Appendix)
♻ ☆ Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation
Many large-scale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show that sparse coding offers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We propose Contrastive Sparse Representation (CSR), a method that sparsifies pre-trained embeddings into a high-dimensional but selectively activated feature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed-often by large margins-while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount. Code is available at https://github.com/neilwen987/CSR_Adaptive_Rep
comment: A novel sparse coding framework designed for learning adaptive representation
♻ ☆ CTC-DRO: Robust Optimization for Reducing Language Disparities in Speech Recognition
Modern deep learning models often achieve high overall performance, but consistently fail on specific subgroups. Group distributionally robust optimization (group DRO) addresses this problem by minimizing the worst-group loss, but it fails when group losses misrepresent performance differences between groups. This is common in domains like speech, where the widely used connectionist temporal classification (CTC) loss scales with input length and varies with linguistic and acoustic properties, leading to spurious differences between group losses. We present CTC-DRO, which addresses the shortcomings of the group DRO objective by smoothing the group weight update to prevent overemphasis on consistently high-loss groups, while using input length-matched batching to mitigate CTC's scaling issues. We evaluate CTC-DRO on the task of multilingual automatic speech recognition (ASR) across five language sets from the ML-SUPERB 2.0 benchmark. CTC-DRO consistently outperforms group DRO and CTC-based baseline models, reducing the worst-language error by up to 47.1% and the average error by up to 32.9%. CTC-DRO can be applied to ASR with minimal computational costs, and offers the potential for reducing group disparities in other domains with similar challenges.
♻ ☆ Bonsai: Gradient-free Graph Distillation for Node Classification
Graph distillation has emerged as a promising avenue to enable scalable training of GNNs by compressing the training dataset while preserving essential graph characteristics. Our study uncovers significant shortcomings in current graph distillation techniques. First, the majority of the algorithms paradoxically require training on the full dataset to perform distillation. Second, due to their gradient-emulating approach, these methods require fresh distillation for any change in hyperparameters or GNN architecture, limiting their flexibility and reusability. Finally, they fail to achieve substantial size reduction due to synthesizing fully-connected, edge-weighted graphs. To address these challenges, we present Bonsai, a novel graph distillation method empowered by the observation that \textit{computation trees} form the fundamental processing units of message-passing GNNs. Bonsai distills datasets by encoding a careful selection of \textit{exemplar} trees that maximize the representation of all computation trees in the training set. This unique approach imparts Bonsai as the first linear-time, model-agnostic graph distillation algorithm for node classification that outperforms existing baselines across $6$ real-world datasets on accuracy, while being $22$ times faster on average. Bonsai is grounded in rigorous mathematical guarantees on the adopted approximation strategies making it robust to GNN architectures, datasets, and parameters.
♻ ☆ Statistical Advantages of Perturbing Cosine Router in Mixture of Experts ICLR 2025
The cosine router in Mixture of Experts (MoE) has recently emerged as an attractive alternative to the conventional linear router. Indeed, the cosine router demonstrates favorable performance in image and language tasks and exhibits better ability to mitigate the representation collapse issue, which often leads to parameter redundancy and limited representation potentials. Despite its empirical success, a comprehensive analysis of the cosine router in MoE has been lacking. Considering the least square estimation of the cosine routing MoE, we demonstrate that due to the intrinsic interaction of the model parameters in the cosine router via some partial differential equations, regardless of the structures of the experts, the estimation rates of experts and model parameters can be as slow as $\mathcal{O}(1/\log^{\tau}(n))$ where $\tau > 0$ is some constant and $n$ is the sample size. Surprisingly, these pessimistic non-polynomial convergence rates can be circumvented by the widely used technique in practice to stabilize the cosine router -- simply adding noises to the $\ell^2$-norms in the cosine router, which we refer to as \textit{perturbed cosine router}. Under the strongly identifiable settings of the expert functions, we prove that the estimation rates for both the experts and model parameters under the perturbed cosine routing MoE are significantly improved to polynomial rates. Finally, we conduct extensive simulation studies in both synthetic and real data settings to empirically validate our theoretical results.
comment: Accepted to ICLR 2025
♻ ☆ What to align in multimodal contrastive learning? ICLR 2025
Humans perceive the world through multisensory integration, blending the information of different modalities to adapt their behavior. Contrastive learning offers an appealing solution for multimodal self-supervised learning. Indeed, by considering each modality as a different view of the same entity, it learns to align features of different modalities in a shared representation space. However, this approach is intrinsically limited as it only learns shared or redundant information between modalities, while multimodal interactions can arise in other ways. In this work, we introduce CoMM, a Contrastive MultiModal learning strategy that enables the communication between modalities in a single multimodal space. Instead of imposing cross- or intra- modality constraints, we propose to align multimodal representations by maximizing the mutual information between augmented versions of these multimodal features. Our theoretical analysis shows that shared, synergistic and unique terms of information naturally emerge from this formulation, allowing us to estimate multimodal interactions beyond redundancy. We test CoMM both in a controlled and in a series of real-world settings: in the former, we demonstrate that CoMM effectively captures redundant, unique and synergistic information between modalities. In the latter, CoMM learns complex multimodal interactions and achieves state-of-the-art results on the seven multimodal benchmarks. Code is available at https://github.com/Duplums/CoMM
comment: ICLR 2025, 25 pages
♻ ☆ CycleResearcher: Improving Automated Research via Automated Review ICLR 2025
The automation of scientific discovery has been a long-standing goal within the research community, driven by the potential to accelerate knowledge creation. While significant progress has been made using commercial large language models (LLMs) as research assistants or idea generators, the possibility of automating the entire research process with open-source LLMs remains largely unexplored. This paper explores the feasibility of using open-source post-trained LLMs as autonomous agents capable of performing the full cycle of automated research and review, from literature review and manuscript preparation to peer review and paper refinement. Our iterative preference training framework consists of CycleResearcher, which conducts research tasks, and CycleReviewer, which simulates the peer review process, providing iterative feedback via reinforcement learning. To train these models, we develop two new datasets, Review-5k and Research-14k, reflecting real-world machine learning research and peer review dynamics. Our results demonstrate that CycleReviewer achieves promising performance with a 26.89\% reduction in mean absolute error (MAE) compared to individual human reviewers in predicting paper scores, indicating the potential of LLMs to effectively assist expert-level research evaluation. In research, the papers generated by the CycleResearcher model achieved a score of 5.36 in simulated peer reviews, showing some competitiveness in terms of simulated review scores compared to the preprint level of 5.24 from human experts, while still having room for improvement compared to the accepted paper level of 5.69. This work represents a significant step toward fully automated scientific inquiry, providing ethical safeguards and exploring AI-driven research capabilities. The code, dataset and model weight are released at https://wengsyx.github.io/Researcher/
comment: Accept in ICLR 2025
♻ ☆ Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification
The ability to process long contexts is crucial for many natural language processing tasks, yet it remains a significant challenge. While substantial progress has been made in enhancing the efficiency of attention mechanisms, there is still a gap in understanding how attention heads function in long-context settings. In this paper, we observe that while certain heads consistently attend to local information only, others swing between attending to local and long-context information depending on the query. This raises the question: can we identify which heads require long-context information to predict the next token accurately? We demonstrate that it's possible to predict which heads are crucial for long-context processing using only local keys. The core idea here is to exploit a simple model for the long-context scores via second moment approximations. These findings unveil simple properties of attention in the context of long sequences, and open the door to potentially significant gains in efficiency.
♻ ☆ Generative Adversarial Networks for High-Dimensional Item Factor Analysis: A Deep Adversarial Learning Algorithm
Advances in deep learning and representation learning have transformed item factor analysis (IFA) in the item response theory (IRT) literature by enabling more efficient and accurate parameter estimation. Variational Autoencoders (VAEs) have been one of the most impactful techniques in modeling high-dimensional latent variables in this context. However, the limited expressiveness of the inference model based on traditional VAEs can still hinder the estimation performance. We introduce Adversarial Variational Bayes (AVB) algorithms as an improvement to VAEs for IFA with improved flexibility and accuracy. By bridging the strengths of VAEs and Generative Adversarial Networks (GANs), AVB incorporates an auxiliary discriminator network to reframe the estimation process as a two-player adversarial game and removes the restrictive assumption of standard normal distributions in the inference model. Theoretically, AVB can achieve similar or higher likelihood compared to VAEs. A further enhanced algorithm, Importance-weighted Adversarial Variational Bayes (IWAVB) is proposed and compared with Importance-weighted Autoencoders (IWAE). In an exploratory analysis of empirical data, IWAVB demonstrated superior expressiveness by achieving a higher likelihood compared to IWAE. In confirmatory analysis with simulated data, IWAVB achieved similar mean-square error results to IWAE while consistently achieving higher likelihoods. When latent variables followed a multimodal distribution, IWAVB outperformed IWAE. With its innovative use of GANs, IWAVB is shown to have the potential to extend IFA to handle large-scale data, facilitating the potential integration of psychometrics and multimodal data analysis.
♻ ☆ MMBind: Unleashing the Potential of Distributed and Heterogeneous Data for Multimodal Learning in IoT
Multimodal sensing systems are increasingly prevalent in various real-world applications. Most existing multimodal learning approaches heavily rely on training with a large amount of synchronized, complete multimodal data. However, such a setting is impractical in real-world IoT sensing applications where data is typically collected by distributed nodes with heterogeneous data modalities, and is also rarely labeled. In this paper, we propose MMBind, a new data binding approach for multimodal learning on distributed and heterogeneous IoT data. The key idea of MMBind is to construct a pseudo-paired multimodal dataset for model training by binding data from disparate sources and incomplete modalities through a sufficiently descriptive shared modality. We also propose a weighted contrastive learning approach to handle domain shifts among disparate data, coupled with an adaptive multimodal learning architecture capable of training models with heterogeneous modality combinations. Evaluations on ten real-world multimodal datasets highlight that MMBind outperforms state-of-the-art baselines under varying degrees of data incompleteness and domain shift, and holds promise for advancing multimodal foundation model training in IoT applications\footnote (The source code is available via https://github.com/nesl/multimodal-bind).
♻ ☆ Exploring Code Language Models for Automated HLS-based Hardware Generation: Benchmark, Infrastructure and Analysis SP
Recent advances in code generation have illuminated the potential of employing large language models (LLMs) for general-purpose programming languages such as Python and C++, opening new opportunities for automating software development and enhancing programmer productivity. The potential of LLMs in software programming has sparked significant interest in exploring automated hardware generation and automation. Although preliminary endeavors have been made to adopt LLMs in generating hardware description languages (HDLs), several challenges persist in this direction. First, the volume of available HDL training data is substantially smaller compared to that for software programming languages. Second, the pre-trained LLMs, mainly tailored for software code, tend to produce HDL designs that are more error-prone. Third, the generation of HDL requires a significantly higher number of tokens compared to software programming, leading to inefficiencies in cost and energy consumption. To tackle these challenges, this paper explores leveraging LLMs to generate High-Level Synthesis (HLS)-based hardware design. Although code generation for domain-specific programming languages is not new in the literature, we aim to provide experimental results, insights, benchmarks, and evaluation infrastructure to investigate the suitability of HLS over low-level HDLs for LLM-assisted hardware design generation. To achieve this, we first finetune pre-trained models for HLS-based hardware generation, using a collected dataset with text prompts and corresponding reference HLS designs. An LLM-assisted framework is then proposed to automate end-to-end hardware code generation, which also investigates the impact of chain-of-thought and feedback loops promoting techniques on HLS-design generation. Limited by the timeframe of this research, we plan to evaluate more advanced reasoning models in the future.
comment: Paper accepted by ASP-DAC'25
♻ ☆ One-Shot Imitation under Mismatched Execution
Human demonstrations as prompts are a powerful way to program robots to do long-horizon manipulation tasks. However, translating these demonstrations into robot-executable actions presents significant challenges due to execution mismatches in movement styles and physical capabilities. Existing methods either depend on human-robot paired data, which is infeasible to scale, or rely heavily on frame-level visual similarities that often break down in practice. To address these challenges, we propose RHyME, a novel framework that automatically aligns human and robot task executions using optimal transport costs. Given long-horizon robot demonstrations, RHyME synthesizes semantically equivalent human videos by retrieving and composing short-horizon human clips. This approach facilitates effective policy training without the need for paired data. RHyME successfully imitates a range of cross-embodiment demonstrators, both in simulation and with a real human hand, achieving over 50\% increase in task success compared to previous methods. We release our code and datasets at https://portal-cornell.github.io/rhyme/.
♻ ☆ MDP Geometry, Normalization and Reward Balancing Solvers AISTATS 2025
We present a new geometric interpretation of Markov Decision Processes (MDPs) with a natural normalization procedure that allows us to adjust the value function at each state without altering the advantage of any action with respect to any policy. This advantage-preserving transformation of the MDP motivates a class of algorithms which we call Reward Balancing, which solve MDPs by iterating through these transformations, until an approximately optimal policy can be trivially found. We provide a convergence analysis of several algorithms in this class, in particular showing that for MDPs for unknown transition probabilities we can improve upon state-of-the-art sample complexity results.
comment: AISTATS 2025 camera-ready version
♻ ☆ Capability-Aware Shared Hypernetworks for Flexible Heterogeneous Multi-Robot Coordination
Recent advances have enabled heterogeneous multi-robot teams to learn complex and effective coordination. However, existing architectural designs that support heterogeneous teams tend to force a trade-off between expressivity and efficiency. Some attempt to encode diverse behaviors within a single shared architecture by appending the input with an ID unique to each robot or robot type. These designs improve sample and parameter efficiency but tend to limit behavioral diversity. Others use a separate policy for each robot, enabling greater diversity at the cost of efficiency and generalization. We view these two designs as ends of a spectrum and explore a middle-ground approach that enables efficient learning of diverse behaviors. Inspired by work in transfer learning and meta RL, and building upon prior work in trait-based task allocation, we propose Capability-Aware Shared Hypernetworks (CASH), a general-purpose soft weight sharing architecture that uses hypernetworks to enable a single architecture to dynamically adapt to each robot and the current context. Intuitively, CASH encodes shared decision making strategies that can be adapted to each robot based on local observations and the robots' individual and collective capabilities (e.g., speed and payload). CASH explicitly captures the impact of capabilities on collective behavior, enabling zero-shot generalization to unseen robots or team compositions. We conducted experiments across four heterogeneous coordination tasks and three learning paradigms (imitation learning, value-based, and policy-gradient RL) using SOTA multi-robot simulation (JaxMARL) and hardware (Robotarium) platforms. Across all conditions, CASH generates appropriately diverse behaviors and outperforms baseline architectures in task performance and sample efficiency during training and zero-shot generalization while utilizing 60%-80% fewer learnable parameters.
comment: 16 pages, 8 figures, equal authorship between Kevin Fu and Shalin Jain
♻ ☆ Dashing for the Golden Snitch: Multi-Drone Time-Optimal Motion Planning with Multi-Agent Reinforcement Learning
Recent innovations in autonomous drones have facilitated time-optimal flight in single-drone configurations, and enhanced maneuverability in multi-drone systems by applying optimal control and learning-based methods. However, few studies have achieved time-optimal motion planning for multi-drone systems, particularly during highly agile maneuvers or in dynamic scenarios. This paper presents a decentralized policy network using multi-agent reinforcement learning for time-optimal multi-drone flight. To strike a balance between flight efficiency and collision avoidance, we introduce a soft collision-free mechanism inspired by optimization-based methods. By customizing PPO in a centralized training, decentralized execution (CTDE) fashion, we unlock higher efficiency and stability in training while ensuring lightweight implementation. Extensive simulations show that, despite slight performance trade-offs compared to single-drone systems, our multi-drone approach maintains near-time-optimal performance with a low collision rate. Real-world experiments validate our method, with two quadrotors using the same network as in simulation achieving a maximum speed of 13.65 m/s and a maximum body rate of 13.4 rad/s in a 5.5 m * 5.5 m * 2.0 m space across various tracks, relying entirely on onboard computation.
comment: v2: 7 pages, 6 figures; terminology corrected, algorithmic and equation descriptions revised, references added
♻ ☆ Beyond Canonicalization: How Tensorial Messages Improve Equivariant Message Passing ICLR 2025
In numerous applications of geometric deep learning, the studied systems exhibit spatial symmetries and it is desirable to enforce these. For the symmetry of global rotations and reflections, this means that the model should be equivariant with respect to the transformations that form the group of $\mathrm O(d)$. While many approaches for equivariant message passing require specialized architectures, including non-standard normalization layers or non-linearities, we here present a framework based on local reference frames ("local canonicalization") which can be integrated with any architecture without restrictions. We enhance equivariant message passing based on local canonicalization by introducing tensorial messages to communicate geometric information consistently between different local coordinate frames. Our framework applies to message passing on geometric data in Euclidean spaces of arbitrary dimension. We explicitly show how our approach can be adapted to make a popular existing point cloud architecture equivariant. We demonstrate the superiority of tensorial messages and achieve state-of-the-art results on normal vector regression and competitive results on other standard 3D point cloud tasks.
comment: To be published in proceedings of ICLR 2025
♻ ☆ Incentivizing Truthful Collaboration in Heterogeneous Federated Learning
Federated learning (FL) is a distributed collaborative learning method, where multiple clients learn together by sharing gradient updates instead of raw data. However, it is well-known that FL is vulnerable to manipulated updates from clients. In this work we study the impact of data heterogeneity on clients' incentives to manipulate their updates. First, we present heterogeneous collaborative learning scenarios where a client can modify their updates to be better off, and show that these manipulations can lead to diminishing model performance. To prevent such modifications, we formulate a game in which clients may misreport their gradient updates in order to "steer" the server model to their advantage. We develop a payment rule that provably disincentivizes sending modified updates under the FedSGD protocol. We derive explicit bounds on the clients' payments and the convergence rate of the global model, which allows us to study the trade-off between heterogeneity, payments and convergence. Finally, we provide an experimental evaluation of the effectiveness of our payment rule in the FedSGD, median-based aggregation FedSGD and FedAvg protocols on three tasks in computer vision and natural language processing. In all cases we find that our scheme successfully disincentivizes modifications.
comment: 29 pages, 8 figures
♻ ☆ On the Utility of Equivariance and Symmetry Breaking in Deep Learning Architectures on Point Clouds
This paper explores the key factors that influence the performance of models working with point clouds, across different tasks of varying geometric complexity. In this work, we explore the trade-offs between flexibility and weight-sharing introduced by equivariant layers, assessing when equivariance boosts or detracts from performance. It is often argued that providing more information as input improves a model's performance. However, if this additional information breaks certain properties, such as $\SE(3)$ equivariance, does it remain beneficial? We identify the key aspects of equivariant and non-equivariant architectures that drive success in different tasks by benchmarking them on segmentation, regression, and generation tasks across multiple datasets with increasing complexity. We observe a positive impact of equivariance, which becomes more pronounced with increasing task complexity, even when strict equivariance is not required.
comment: 19 pages, 4 figures
♻ ☆ Efficient Neural SDE Training using Wiener-Space Cubature
A neural stochastic differential equation (SDE) is an SDE with drift and diffusion terms parametrized by neural networks. The training procedure for neural SDEs consists of optimizing the SDE vector field (neural network) parameters to minimize the expected value of an objective functional on infinite-dimensional path-space. Existing training techniques focus on methods to efficiently compute path-wise gradients of the objective functional with respect to these parameters, then pair this with Monte-Carlo simulation to estimate the expectation, and stochastic gradient descent to optimize. In this work we introduce a novel training technique which bypasses and improves upon Monte-Carlo simulation; we extend results in the theory of Wiener-space cubature to approximate the expected objective functional by a weighted sum of deterministic ODE solutions. This allows us to compute gradients by efficient ODE adjoint methods. Furthermore, we exploit a high-order recombination scheme to drastically reduce the number of ODE solutions necessary to achieve a reasonable approximation. We show that this Wiener-space cubature approach can surpass the O(1/sqrt(n)) rate of Monte-Carlo simulation, or the O(log(n)/n) rate of quasi-Monte-Carlo, to achieve a O(1/n) rate under reasonable assumptions.
♻ ☆ LLMs can be Dangerous Reasoners: Analyzing-based Jailbreak Attack on Large Language Models
The rapid development of Large Language Models (LLMs) has brought significant advancements across various tasks. However, despite these achievements, LLMs still exhibit inherent safety vulnerabilities, especially when confronted with jailbreak attacks. Existing jailbreak methods suffer from two main limitations: reliance on complicated prompt engineering and iterative optimization, which lead to low attack success rate (ASR) and attack efficiency (AE). In this work, we propose an efficient jailbreak attack method, Analyzing-based Jailbreak (ABJ), which leverages the advanced reasoning capability of LLMs to autonomously generate harmful content, revealing their underlying safety vulnerabilities during complex reasoning process. We conduct comprehensive experiments on ABJ across various open-source and closed-source LLMs. In particular, ABJ achieves high ASR (82.1% on GPT-4o-2024-11-20) with exceptional AE among all target LLMs, showcasing its remarkable attack effectiveness, transferability, and efficiency. Our findings underscore the urgent need to prioritize and improve the safety of LLMs to mitigate the risks of misuse.
♻ ☆ Online Scheduling for LLM Inference with KV Cache Constraints
Large Language Model (LLM) inference, where a trained model generates text one word at a time in response to user prompts, is a computationally intensive process requiring efficient scheduling to optimize latency and resource utilization. A key challenge in LLM inference is the management of the Key-Value (KV) cache, which reduces redundant computations but introduces memory constraints. In this work, we model LLM inference with KV cache constraints theoretically and propose novel batching and scheduling algorithms that minimize inference latency while effectively managing the KV cache's memory. We analyze both semi-online and fully online scheduling models, and our results are threefold. First, we provide a polynomial-time algorithm that achieves exact optimality in terms of average latency in the semi-online prompt arrival model. Second, in the fully online case with a stochastic prompt arrival, we introduce an efficient online scheduling algorithm with constant regret. Third, we prove that no algorithm (deterministic or randomized) can achieve a constant competitive ratio in fully online adversarial settings. Our empirical evaluations on a public LLM inference dataset, using the Llama-70B model on A100 GPUs, show that our approach significantly outperforms benchmark algorithms used currently in practice, achieving lower latency while reducing energy consumption. Overall, our results offer a path toward more sustainable and cost-effective LLM deployment.
comment: Will add a lemma in the proof of Theorem 5.3 to make the statement and proof more rigorous
♻ ☆ GSplatLoc: Grounding Keypoint Descriptors into 3D Gaussian Splatting for Improved Visual Localization
Although various visual localization approaches exist, such as scene coordinate regression and camera pose regression, these methods often struggle with optimization complexity or limited accuracy. To address these challenges, we explore the use of novel view synthesis techniques, particularly 3D Gaussian Splatting (3DGS), which enables the compact encoding of both 3D geometry and scene appearance. We propose a two-stage procedure that integrates dense and robust keypoint descriptors from the lightweight XFeat feature extractor into 3DGS, enhancing performance in both indoor and outdoor environments. The coarse pose estimates are directly obtained via 2D-3D correspondences between the 3DGS representation and query image descriptors. In the second stage, the initial pose estimate is refined by minimizing the rendering-based photometric warp loss. Benchmarking on widely used indoor and outdoor datasets demonstrates improvements over recent neural rendering-based localization methods, such as NeRFMatch and PNeRFLoc.
comment: Project website at https://gsplatloc.github.io/
♻ ☆ PCM Selector: Penalized Covariate-Mediator Selection Operator for Evaluating Linear Causal Effects AAAI 2025
For a data-generating process for random variables that can be described with a linear structural equation model, we consider a situation in which (i) a set of covariates satisfying the back-door criterion cannot be observed or (ii) such a set can be observed, but standard statistical estimation methods cannot be applied to estimate causal effects because of multicollinearity/high-dimensional data problems. We propose a novel two-stage penalized regression approach, the penalized covariate-mediator selection operator (PCM Selector), to estimate the causal effects in such scenarios. Unlike existing penalized regression analyses, when a set of intermediate variables is available, PCM Selector provides a consistent or less biased estimator of the causal effect. In addition, PCM Selector provides a variable selection procedure for intermediate variables to obtain better estimation accuracy of the causal effects than does the back-door criterion.
comment: Accepted by AAAI 2025
♻ ☆ How simple can you go? An off-the-shelf transformer approach to molecular dynamics
Most current neural networks for molecular dynamics (MD) include physical inductive biases, resulting in specialized and complex architectures. This is in contrast to most other machine learning domains, where specialist approaches are increasingly replaced by general-purpose architectures trained on vast datasets. In line with this trend, several recent studies have questioned the necessity of architectural features commonly found in MD models, such as built-in rotational equivariance or energy conservation. In this work, we contribute to the ongoing discussion by evaluating the performance of an MD model with as few specialized architectural features as possible. We present a recipe for MD using an Edge Transformer, an "off-the-shelf'' transformer architecture that has been minimally modified for the MD domain, termed MD-ET. Our model implements neither built-in equivariance nor energy conservation. We use a simple supervised pre-training scheme on $\sim$30 million molecular structures from the QCML database. Using this "off-the-shelf'' approach, we show state-of-the-art results on several benchmarks after fine-tuning for a small number of steps. Additionally, we examine the effects of being only approximately equivariant and energy conserving for MD simulations, proposing a novel method for distinguishing the errors resulting from non-equivariance from other sources of inaccuracies like numerical rounding errors. While our model exhibits runaway energy increases on larger structures, we show approximately energy-conserving NVE simulations for a range of small structures.
comment: 21 pages, code at https://github.com/mx-e/simple-md
♻ ☆ From Sparse Dependence to Sparse Attention: Unveiling How Chain-of-Thought Enhances Transformer Sample Efficiency
Chain-of-thought (CoT) significantly enhances the reasoning performance of large language models (LLM). While current theoretical studies often attribute this improvement to increased expressiveness and computational capacity, we argue that expressiveness is not the primary limitation in the LLM regime, as current large models will fail on simple tasks. Using a parity-learning setup, we demonstrate that CoT can substantially improve sample efficiency even when the representation power is sufficient. Specifically, with CoT, a transformer can learn the function within polynomial samples, whereas without CoT, the required sample size is exponential. Additionally, we show that CoT simplifies the learning process by introducing sparse sequential dependencies among input tokens, and leads to a sparse and interpretable attention. We validate our theoretical analysis with both synthetic and real-world experiments, confirming that sparsity in attention layers is a key factor of the improvement induced by CoT.
comment: 43 pages,11 figures
♻ ☆ Handling Spatial-Temporal Data Heterogeneity for Federated Continual Learning via Tail Anchor CVPR 2025
Federated continual learning (FCL) allows each client to continually update its knowledge from task streams, enhancing the applicability of federated learning in real-world scenarios. However, FCL needs to address not only spatial data heterogeneity between clients but also temporal data heterogeneity between tasks. In this paper, empirical experiments demonstrate that such input-level heterogeneity significantly affects the model's internal parameters and outputs, leading to severe spatial-temporal catastrophic forgetting of local and previous knowledge. To this end, we propose Federated Tail Anchor (FedTA) to mix trainable Tail Anchor with the frozen output features to adjust their position in the feature space, thereby overcoming parameter-forgetting and output-forgetting. Three novel components are also included: Input Enhancement for improving the performance of pre-trained models on downstream tasks; Selective Input Knowledge Fusion for fusion of heterogeneous local knowledge on the server; and Best Global Prototype Selection for finding the best anchor point for each class in the feature space. Extensive experiments demonstrate that FedTA not only outperforms existing FCL methods but also effectively preserves the relative positions of features.
comment: This paper is accepted by CVPR 2025
♻ ☆ Promote, Suppress, Iterate: How Language Models Answer One-to-Many Factual Queries
To answer one-to-many factual queries (e.g., listing cities of a country), a language model (LM) must simultaneously recall knowledge and avoid repeating previous answers. How are these two subtasks implemented and integrated internally? Across multiple datasets and models, we identify a promote-then-suppress mechanism: the model first recalls all answers, and then suppresses previously generated ones. Specifically, LMs use both the subject and previous answer tokens to perform knowledge recall, with attention propagating subject information and MLPs promoting the answers. Then, attention attends to and suppresses previous answer tokens, while MLPs amplify the suppression signal. Our mechanism is corroborated by extensive experimental evidence: in addition to using early decoding and causal tracing, we analyze how components use different tokens by introducing both Token Lens, which decodes aggregated attention updates from specified tokens, and a knockout method that analyzes changes in MLP outputs after removing attention to specified tokens. Overall, we provide new insights into how LMs' internal components interact with different input tokens to support complex factual recall. Code is available at https://github.com/Lorenayannnnn/how-lms-answer-one-to-many-factual-queries.
♻ ☆ Graph-Aware Isomorphic Attention for Adaptive Dynamics in Transformers
We present an approach to modifying Transformer architectures by integrating graph-aware relational reasoning into the attention mechanism, merging concepts from graph neural networks and language modeling. Building on the inherent connection between attention and graph theory, we reformulate the Transformer's attention mechanism as a graph operation and propose Graph-Aware Isomorphic Attention. This method leverages advanced graph modeling strategies, including Graph Isomorphism Networks (GIN) and Principal Neighborhood Aggregation (PNA), to enrich the representation of relational structures. Our approach captures complex dependencies and generalizes across tasks, as evidenced by a reduced generalization gap and improved learning performance. Additionally, we expand the concept of graph-aware attention to introduce Sparse GIN-Attention, a fine-tuning approach that employs sparse GINs. By interpreting attention matrices as sparse adjacency graphs, this technique enhances the adaptability of pre-trained foundational models with minimal computational overhead, endowing them with graph-aware capabilities. Sparse GIN-Attention fine-tuning achieves improved training dynamics and better generalization compared to alternative methods like low-rank adaption (LoRA). We discuss latent graph-like structures within traditional attention mechanisms, offering a new lens through which Transformers can be understood. By evolving Transformers as hierarchical GIN models for relational reasoning. This perspective suggests profound implications for foundational model development, enabling the design of architectures that dynamically adapt to both local and global dependencies. Applications in bioinformatics, materials science, language modeling, and beyond could benefit from this synthesis of relational and sequential data modeling, setting the stage for interpretable and generalizable modeling strategies.
♻ ☆ Zero-Knowledge Proof-based Verifiable Decentralized Machine Learning in Communication Network: A Comprehensive Survey
Over recent decades, machine learning has significantly advanced network communication, enabling improved decision-making, user behavior analysis, and fault detection. Decentralized approaches, where participants exchange computation results instead of raw private data, mitigate these risks but introduce challenges related to trust and verifiability. A critical issue arises: How can one ensure the integrity and validity of computation results shared by other participants? Existing survey articles predominantly address security and privacy concerns in decentralized machine learning, whereas this survey uniquely highlights the emerging issue of verifiability. Recognizing the critical role of zero-knowledge proofs in ensuring verifiability, we present a comprehensive review of Zero-Knowledge Proof-based Verifiable Machine Learning (ZKP-VML). To clarify the research problem, we present a definition of ZKP-VML consisting of four algorithms, along with several corresponding key security properties. Besides, we provide an overview of the current research landscape by systematically organizing the research timeline and categorizing existing schemes based on their security properties. Furthermore, through an in-depth analysis of each existing scheme, we summarize their technical contributions and optimization strategies, aiming to uncover common design principles underlying ZKP-VML schemes. Building on the reviews and analysis presented, we identify current research challenges and suggest future research directions. To the best of our knowledge, this is the most comprehensive survey to date on verifiable decentralized machine learning and ZKP-VML.
♻ ☆ Channel-Attentive Graph Neural Networks
Graph Neural Networks (GNNs) set the state-of-the-art in representation learning for graph-structured data. They are used in many domains, from online social networks to complex molecules. Most GNNs leverage the message-passing paradigm and achieve strong performances on various tasks. However, the message-passing mechanism used in most models suffers from over-smoothing as a GNN's depth increases. The over-smoothing degrades GNN's performance due to the increased similarity between the representations of unrelated nodes. This study proposes an adaptive channel-wise message-passing approach to alleviate the over-smoothing. The proposed model, Channel-Attentive GNN, learns how to attend to neighboring nodes and their feature channels. Thus, much diverse information can be transferred between nodes during message-passing. Experiments with widely used benchmark datasets show that the proposed model is more resistant to over-smoothing than baselines and achieves state-of-the-art performances for various graphs with strong heterophily. Our code is at https://github.com/ALLab-Boun/CHAT-GNN.
comment: Published as a conference paper at IEEE International Conference on Data Mining 2024
♻ ☆ LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
We introduce LADDER (Learning through Autonomous Difficulty-Driven Example Recursion), a framework which enables Large Language Models to autonomously improve their problem-solving capabilities through self-guided learning by recursively generating and solving progressively simpler variants of complex problems. Unlike prior approaches that require curated datasets or human feedback, LADDER leverages a model's own capabilities to generate easier question variants. We demonstrate LADDER's effectiveness in the subject of mathematical integration, improving Llama 3.2 3B's accuracy from 1% to 82% on undergraduate-level problems and enabling Qwen2.5 7B Deepseek-R1 Distilled to achieve 73% on the MIT Integration Bee qualifying examination. We also introduce TTRL (Test-Time Reinforcement Learning), where we perform reinforcement learning on variants of test problems at inference time. TTRL enables Qwen2.5 7B Deepseek-R1 Distilled to achieve a state-of-the-art score of 90% on the MIT Integration Bee qualifying examination, surpassing OpenAI o1's performance. These results show how self-directed strategic learning can achieve significant capability improvements without relying on architectural scaling or human supervision.
♻ ☆ ChaI-TeA: A Benchmark for Evaluating Autocompletion of Interactions with LLM-based Chatbots
The rise of LLMs has deflected a growing portion of human-computer interactions towards LLM-based chatbots. The remarkable abilities of these models allow users to interact using long, diverse natural language text covering a wide range of topics and styles. Phrasing these messages is a time and effort consuming task, calling for an autocomplete solution to assist users. We introduce the task of chatbot interaction autocomplete. We present ChaI-TeA: CHat InTEraction Autocomplete; An autcomplete evaluation framework for LLM-based chatbot interactions. The framework includes a formal definition of the task, coupled with suitable datasets and metrics. We use the framework to evaluate After formally defining the task along with suitable datasets and metrics, we test 9 models on the defined auto completion task, finding that while current off-the-shelf models perform fairly, there is still much room for improvement, mainly in ranking of the generated suggestions. We provide insights for practitioners working on this task and open new research directions for researchers in the field. We release our framework to serve as a foundation for future research.
♻ ☆ AIArena: A Blockchain-Based Decentralized AI Training Platform WWW
The rapid advancement of AI has underscored critical challenges in its development and implementation, largely due to centralized control by a few major corporations. This concentration of power intensifies biases within AI models, resulting from inadequate governance and oversight mechanisms. Additionally, it limits public involvement and heightens concerns about the integrity of model generation. Such monopolistic control over data and AI outputs threatens both innovation and fair data usage, as users inadvertently contribute data that primarily benefits these corporations. In this work, we propose AIArena, a blockchain-based decentralized AI training platform designed to democratize AI development and alignment through on-chain incentive mechanisms. AIArena fosters an open and collaborative environment where participants can contribute models and computing resources. Its on-chain consensus mechanism ensures fair rewards for participants based on their contributions. We instantiate and implement AIArena on the public Base blockchain Sepolia testnet, and the evaluation results demonstrate the feasibility of AIArena in real-world applications.
comment: Camera ready version. Accepted by the ACM Web Conference (WWW), 2025
♻ ☆ Mitigating the Stability-Plasticity Dilemma in Adaptive Train Scheduling with Curriculum-Driven Continual DQN Expansion
A continual learning agent builds on previous experiences to develop increasingly complex behaviors by adapting to non-stationary and dynamic environments while preserving previously acquired knowledge. However, scaling these systems presents significant challenges, particularly in balancing the preservation of previous policies with the adaptation of new ones to current environments. This balance, known as the stability-plasticity dilemma, is especially pronounced in complex multi-agent domains such as the train scheduling problem, where environmental and agent behaviors are constantly changing, and the search space is vast. In this work, we propose addressing these challenges in the train scheduling problem using curriculum learning. We design a curriculum with adjacent skills that build on each other to improve generalization performance. Introducing a curriculum with distinct tasks introduces non-stationarity, which we address by proposing a new algorithm: Continual Deep Q-Network (DQN) Expansion (CDE). Our approach dynamically generates and adjusts Q-function subspaces to handle environmental changes and task requirements. CDE mitigates catastrophic forgetting through EWC while ensuring high plasticity using adaptive rational activation functions. Experimental results demonstrate significant improvements in learning efficiency and adaptability compared to RL baselines and other adapted methods for continual learning, highlighting the potential of our method in managing the stability-plasticity dilemma in the adaptive train scheduling setting.
comment: 9 Pages, 2 Figures
♻ ☆ UniFlow: A Foundation Model for Unified Urban Spatio-Temporal Flow Prediction
Urban spatio-temporal flow prediction, encompassing traffic flows and crowd flows, is crucial for optimizing city infrastructure and managing traffic and emergency responses. Traditional approaches have relied on separate models tailored to either grid-based data, representing cities as uniform cells, or graph-based data, modeling cities as networks of nodes and edges. In this paper, we build UniFlow, a foundational model for general urban flow prediction that unifies both grid-based and graphbased data. We first design a multi-view spatio-temporal patching mechanism to standardize different data into a consistent sequential format and then introduce a spatio-temporal transformer architecture to capture complex correlations and dynamics. To leverage shared spatio-temporal patterns across different data types and facilitate effective cross-learning, we propose SpatioTemporal Memory Retrieval Augmentation (ST-MRA). By creating structured memory modules to store shared spatio-temporal patterns, ST-MRA enhances predictions through adaptive memory retrieval. Extensive experiments demonstrate that UniFlow outperforms existing models in both grid-based and graph-based flow prediction, excelling particularly in scenarios with limited data availability, showcasing its superior performance and broad applicability. The datasets and code implementation have been released on https://github.com/YuanYuan98/UniFlow.
♻ ☆ CodeIF: Benchmarking the Instruction-Following Capabilities of Large Language Models for Code Generation
With the rapid advancement of Large Language Models (LLMs), the demand for robust instruction-following capabilities in code generation tasks has grown significantly. Code generation not only facilitates faster prototyping and automated testing, but also augments developer efficiency through improved maintainability and reusability of code. In this paper, we introduce CodeIF, the first benchmark specifically designed to assess the abilities of LLMs to adhere to task-oriented instructions within diverse code generation scenarios. CodeIF encompasses a broad range of tasks, including function synthesis, error debugging, algorithmic refactoring, and code explanation, thereby providing a comprehensive suite to evaluate model performance across varying complexity levels and programming domains. We conduct extensive experiments with LLMs, analyzing their strengths and limitations in meeting the demands of these tasks. The experimental results offer valuable insights into how well current models align with human instructions, as well as the extent to which they can generate consistent, maintainable, and contextually relevant code. Our findings not only underscore the critical role that instruction-following LLMs can play in modern software development, but also illuminate pathways for future research aimed at enhancing their adaptability, reliability, and overall effectiveness in automated code generation.
♻ ☆ Bounding Evidence and Estimating Log-Likelihood in VAE AISTATS 2023
Many crucial problems in deep learning and statistical inference are caused by a variational gap, i.e., a difference between model evidence (log-likelihood) and evidence lower bound (ELBO). In particular, in a classical VAE setting that involves training via an ELBO cost function, it is difficult to provide a robust comparison of the effects of training between models, since we do not know a log-likelihood of data (but only its lower bound). In this paper, to deal with this problem, we introduce a general and effective upper bound, which allows us to efficiently approximate the evidence of data. We provide extensive theoretical and experimental studies of our approach, including its comparison to the other state-of-the-art upper bounds, as well as its application as a tool for the evaluation of models that were trained on various lower bounds.
comment: Paper accepted for AISTATS 2023
♻ ☆ DrugAgent: Automating AI-aided Drug Discovery Programming through LLM Multi-Agent Collaboration
Recent progress in Large Language Models (LLMs) has drawn attention to their potential for accelerating drug discovery. However, a central problem remains: translating theoretical ideas into robust implementations in the highly specialized context of pharmaceutical research. This limitation prevents practitioners from making full use of the latest AI developments in drug discovery. To address this challenge, we introduce DrugAgent, a multi-agent framework that automates machine learning (ML) programming for drug discovery tasks. DrugAgent employs an LLM Planner that formulates high-level ideas and an LLM Instructor that identifies and integrates domain knowledge when implementing those ideas. We present case studies on three representative drug discovery tasks. Our results show that DrugAgent consistently outperforms leading baselines, including a relative improvement of 4.92% in ROC-AUC compared to ReAct for drug-target interaction (DTI). DrugAgent is publicly available at https://anonymous.4open.science/r/drugagent-5C42/.
♻ ☆ TAG: A Decentralized Framework for Multi-Agent Hierarchical Reinforcement Learning
Hierarchical organization is fundamental to biological systems and human societies, yet artificial intelligence systems often rely on monolithic architectures that limit adaptability and scalability. Current hierarchical reinforcement learning (HRL) approaches typically restrict hierarchies to two levels or require centralized training, which limits their practical applicability. We introduce TAME Agent Framework (TAG), a framework for constructing fully decentralized hierarchical multi-agent systems. TAG enables hierarchies of arbitrary depth through a novel LevelEnv concept, which abstracts each hierarchy level as the environment for the agents above it. This approach standardizes information flow between levels while preserving loose coupling, allowing for seamless integration of diverse agent types. We demonstrate the effectiveness of TAG by implementing hierarchical architectures that combine different RL agents across multiple levels, achieving improved performance over classical multi-agent RL baselines on standard benchmarks. Our results show that decentralized hierarchical organization enhances both learning speed and final performance, positioning TAG as a promising direction for scalable multi-agent systems.
♻ ☆ Exploration Implies Data Augmentation: Reachability and Generalisation in Contextual MDPs
In the zero-shot policy transfer (ZSPT) setting for contextual Markov decision processes (MDP), agents train on a fixed set of contexts and must generalise to new ones. Recent work has argued and demonstrated that increased exploration can improve this generalisation, by training on more states in the training contexts. In this paper, we demonstrate that training on more states can indeed improve generalisation, but can come at a cost of reducing the accuracy of the learned value function which should not benefit generalisation. We introduce reachability in the ZSPT setting to define which states/contexts require generalisation and explain why exploration can improve it. We hypothesise and demonstrate that using exploration to increase the agent's coverage while also increasing the accuracy improves generalisation even more. Inspired by this, we propose a method Explore-Go that implements an exploration phase at the beginning of each episode, which can be combined with existing on- and off-policy RL algorithms and significantly improves generalisation even in partially observable MDPs. We demonstrate the effectiveness of Explore-Go when combined with several popular algorithms and show an increase in generalisation performance across several environments. With this, we hope to provide practitioners with a simple modification that can improve the generalisation of their agents.
comment: arXiv admin note: text overlap with arXiv:2406.08069
♻ ☆ Unifying Causal Representation Learning with the Invariance Principle ICLR2025
Causal representation learning (CRL) aims at recovering latent causal variables from high-dimensional observations to solve causal downstream tasks, such as predicting the effect of new interventions or more robust classification. A plethora of methods have been developed, each tackling carefully crafted problem settings that lead to different types of identifiability. These different settings are widely assumed to be important because they are often linked to different rungs of Pearl's causal hierarchy, even though this correspondence is not always exact. This work shows that instead of strictly conforming to this hierarchical mapping, many causal representation learning approaches methodologically align their representations with inherent data symmetries. Identification of causal variables is guided by invariance principles that are not necessarily causal. This result allows us to unify many existing approaches in a single method that can mix and match different assumptions, including non-causal ones, based on the invariance relevant to the problem at hand. It also significantly benefits applicability, which we demonstrate by improving treatment effect estimation on real-world high-dimensional ecological data. Overall, this paper clarifies the role of causal assumptions in the discovery of causal variables and shifts the focus to preserving data symmetries.
comment: ICLR2025 Camera ready
♻ ☆ From Learning to Optimize to Learning Optimization Algorithms AISTATS 2025
Towards designing learned optimization algorithms that are usable beyond their training setting, we identify key principles that classical algorithms obey, but have up to now, not been used for Learning to Optimize (L2O). Following these principles, we provide a general design pipeline, taking into account data, architecture and learning strategy, and thereby enabling a synergy between classical optimization and L2O, resulting in a philosophy of Learning Optimization Algorithms. As a consequence our learned algorithms perform well far beyond problems from the training distribution. We demonstrate the success of these novel principles by designing a new learning-enhanced BFGS algorithm and provide numerical experiments evidencing its adaptation to many settings at test time.
comment: To appear at AISTATS 2025
♻ ☆ State Space Models are Provably Comparable to Transformers in Dynamic Token Selection
Deep neural networks based on state space models (SSMs) are attracting significant attention in sequence modeling since their computational cost is much smaller than that of Transformers. While the capabilities of SSMs have been demonstrated through experiments in various tasks, theoretical understanding of SSMs is still limited. In particular, most theoretical studies discuss the capabilities of SSM layers without nonlinear layers, and there is a lack of discussion on their combination with nonlinear layers. In this paper, we explore the capabilities of SSMs combined with fully connected neural networks, and show that they are comparable to Transformers in extracting the essential tokens depending on the input. As concrete examples, we consider two synthetic tasks, which are challenging for a single SSM layer, and demonstrate that SSMs combined with nonlinear layers can efficiently solve these tasks. Furthermore, we study the nonparametric regression task, and prove that the ability of SSMs is equivalent to that of Transformers in estimating functions belonging to a certain class.
comment: 43 pages, 7 figures
♻ ☆ Narrowing the Gap between Adversarial and Stochastic MDPs via Policy Optimization
We consider the problem of learning in adversarial Markov decision processes [MDPs] with an oblivious adversary in a full-information setting. The agent interacts with an environment during $T$ episodes, each of which consists of $H$ stages, and each episode is evaluated with respect to a reward function that will be revealed only at the end of the episode. We propose an algorithm, called APO-MVP, that achieves a regret bound of order $\tilde{\mathcal{O}}(\mathrm{poly}(H)\sqrt{SAT})$, where $S$ and $A$ are sizes of the state and action spaces, respectively. This result improves upon the best-known regret bound by a factor of $\sqrt{S}$, bridging the gap between adversarial and stochastic MDPs, and matching the minimax lower bound $\Omega(\sqrt{H^3SAT})$ as far as the dependencies in $S,A,T$ are concerned. The proposed algorithm and analysis completely avoid the typical tool given by occupancy measures; instead, it performs policy optimization based only on dynamic programming and on a black-box online linear optimization strategy run over estimated advantage functions, making it easy to implement. The analysis leverages two recent techniques: policy optimization based on online linear optimization strategies (Jonckheere et al., 2023) and a refined martingale analysis of the impact on values of estimating transitions kernels (Zhang et al., 2023).
♻ ☆ Distributed Differentially Private Data Analytics via Secure Sketching
We introduce the linear-transformation model, a distributed model of differentially private data analysis. Clients have access to a trusted platform capable of applying a public matrix to their inputs. Such computations can be securely distributed across multiple servers using simple and efficient secure multiparty computation techniques. The linear-transformation model serves as an intermediate model between the highly expressive central model and the minimal local model. In the central model, clients have access to a trusted platform capable of applying any function to their inputs. However, this expressiveness comes at a cost, as it is often prohibitively expensive to distribute such computations, leading to the central model typically being implemented by a single trusted server. In contrast, the local model assumes no trusted platform, which forces clients to add significant noise to their data. The linear-transformation model avoids the single point of failure for privacy present in the central model, while also mitigating the high noise required in the local model. We demonstrate that linear transformations are very useful for differential privacy, allowing for the computation of linear sketches of input data. These sketches largely preserve utility for tasks such as private low-rank approximation and private ridge regression, while introducing only minimal error, critically independent of the number of clients.
♻ ☆ Task-optimal data-driven surrogate models for eNMPC via differentiable simulation and optimization
Mechanistic dynamic process models may be too computationally expensive to be usable as part of a real-time capable predictive controller. We present a method for end-to-end learning of Koopman surrogate models for optimal performance in a specific control task. In contrast to previous contributions that employ standard reinforcement learning (RL) algorithms, we use a training algorithm that exploits the differentiability of environments based on mechanistic simulation models to aid the policy optimization. We evaluate the performance of our method by comparing it to that of other training algorithms on an existing economic nonlinear model predictive control (eNMPC) case study of a continuous stirred-tank reactor (CSTR) model. Compared to the benchmark methods, our method produces similar economic performance while eliminating constraint violations. Thus, for this case study, our method outperforms the others and offers a promising path toward more performant controllers that employ dynamic surrogate models.
comment: 8 pages, 4 figures, 1 table
♻ ☆ DarwinLM: Evolutionary Structured Pruning of Large Language Models
Large Language Models (LLMs) have achieved significant success across various NLP tasks. However, their massive computational costs limit their widespread use, particularly in real-time applications. Structured pruning offers an effective solution by compressing models and directly providing end-to-end speed improvements, regardless of the hardware environment. Meanwhile, different components of the model exhibit varying sensitivities towards pruning, calling for non-uniform model compression. However, a pruning method should not only identify a capable substructure, but also account for post-compression training. To this end, we propose DarwinLM, a method for training-aware structured pruning. DarwinLM builds upon an evolutionary search process, generating multiple offspring models in each generation through mutation, and selecting the fittest for survival. To assess the effect of post-training, we incorporate a lightweight, multistep training process within the offspring population, progressively increasing the number of tokens and eliminating poorly performing models in each selection stage. We validate our method through extensive experiments on Llama-2-7B, Llama-3.1-8B and Qwen-2.5-14B-Instruct, achieving state-of-the-art performance for structured pruning. For instance, DarwinLM surpasses ShearedLlama while requiring 5x less training data during post-compression training. Code is at: https://github.com/IST-DASLab/DarwinLM
comment: Code: https://github.com/IST-DASLab/DarwinLM
♻ ☆ Verifiable and Provably Secure Machine Unlearning
Machine unlearning aims to remove points from the training dataset of a machine learning model after training: e.g., when a user requests their data to be deleted. While many unlearning methods have been proposed, none of them enable users to audit the procedure. Furthermore, recent work shows a user is unable to verify whether their data was unlearnt from an inspection of the model parameter alone. Rather than reasoning about parameters, we propose to view verifiable unlearning as a security problem. To this end, we present the first cryptographic definition of verifiable unlearning to formally capture the guarantees of an unlearning system. In this framework, the server first computes a proof that the model was trained on a dataset D. Given a user's data point d requested to be deleted, the server updates the model using an unlearning algorithm. It then provides a proof of the correct execution of unlearning and that d is not part of D', where D' is the new training dataset (i.e., d has been removed). Our framework is generally applicable to different unlearning techniques that we abstract as admissible functions. We instantiate a protocol in the framework, based on cryptographic assumptions, using SNARKs and hash chains. Finally, we implement the protocol for three different unlearning techniques and validate its feasibility for linear regression, logistic regression, and neural networks.
comment: Accepted at IEEE SaTML2025
♻ ☆ Solving Inverse Problem for Multi-armed Bandits via Convex Optimization
We consider the inverse problem of multi-armed bandits (IMAB) that are widely used in neuroscience and psychology research for behavior modelling. We first show that the IMAB problem is not convex in general, but can be relaxed to a convex problem via variable transformation. Based on this result, we propose a two-step sequential heuristic for (approximately) solving the IMAB problem. We discuss a condition where our method provides global solution to the IMAB problem with certificate, as well as approximations to further save computing time. Numerical experiments indicate that our heuristic method is more robust than directly solving the IMAB problem via repeated local optimization, and can achieve the performance of Monte Carlo methods within a significantly decreased running time. We provide the implementation of our method based on CVXPY, which allows straightforward application by users not well versed in convex optimization.
♻ ☆ POMDP-Driven Cognitive Massive MIMO Radar: Joint Target Detection-Tracking In Unknown Disturbances
The joint detection and tracking of a moving target embedded in an unknown disturbance represents a key feature that motivates the development of the cognitive radar paradigm. Building upon recent advancements in robust target detection with multiple-input multiple-output (MIMO) radars, this work explores the application of a Partially Observable Markov Decision Process (POMDP) framework to enhance the tracking and detection tasks in a statistically unknown environment. In the POMDP setup, the radar system is considered as an intelligent agent that continuously senses the surrounding environment, optimizing its actions to maximize the probability of detection $(P_D)$ and improve the target position and velocity estimation, all this while keeping a constant probability of false alarm $(P_{FA})$. The proposed approach employs an online algorithm that does not require any apriori knowledge of the noise statistics, and it relies on a much more general observation model than the traditional range-azimuth-elevation model employed by conventional tracking algorithms. Simulation results clearly show substantial performance improvement of the POMDP-based algorithm compared to the State-Action-Reward-State-Action (SARSA)-based one that has been recently investigated in the context of massive MIMO (MMIMO) radar systems.
comment: The paper has been submitted to ieee Transactions on radar systems
♻ ☆ Gated Delta Networks: Improving Mamba2 with Delta Rule ICLR 2025
Linear Transformers have gained attention as efficient alternatives to standard Transformers, but their performance in retrieval and long-context tasks has been limited. To address these limitations, recent work has explored two distinct mechanisms: gating for adaptive memory control and the delta update rule for precise memory modifications. We observe that these mechanisms are complementary: gating enables rapid memory erasure while the delta rule facilitates targeted updates. Building on this insight, we introduce the gated delta rule and develop a parallel training algorithm optimized for modern hardware. Our proposed architecture, Gated DeltaNet, consistently surpasses existing models like Mamba2 and DeltaNet across multiple benchmarks, including language modeling, common-sense reasoning, in-context retrieval, length extrapolation, and long-context understanding. We further enhance performance by developing hybrid architectures that combine Gated DeltaNet layers with sliding window attention or Mamba2 layers, achieving both improved training efficiency and superior task performance.
comment: ICLR 2025 camera ready
♻ ☆ ChemVLM: Exploring the Power of Multimodal Large Language Models in Chemistry Area
Large Language Models (LLMs) have achieved remarkable success and have been applied across various scientific fields, including chemistry. However, many chemical tasks require the processing of visual information, which cannot be successfully handled by existing chemical LLMs. This brings a growing need for models capable of integrating multimodal information in the chemical domain. In this paper, we introduce \textbf{ChemVLM}, an open-source chemical multimodal large language model specifically designed for chemical applications. ChemVLM is trained on a carefully curated bilingual multimodal dataset that enhances its ability to understand both textual and visual chemical information, including molecular structures, reactions, and chemistry examination questions. We develop three datasets for comprehensive evaluation, tailored to Chemical Optical Character Recognition (OCR), Multimodal Chemical Reasoning (MMCR), and Multimodal Molecule Understanding tasks. We benchmark ChemVLM against a range of open-source and proprietary multimodal large language models on various tasks. Experimental results demonstrate that ChemVLM achieves competitive performance across all evaluated tasks. Our model can be found at https://huggingface.co/AI4Chem/ChemVLM-26B.
comment: 11 pages, updated version
♻ ☆ Flow-based Bayesian filtering for high-dimensional nonlinear stochastic dynamical systems
Bayesian filtering for high-dimensional nonlinear stochastic dynamical systems is a fundamental yet challenging problem in many fields of science and engineering. Existing methods face significant obstacles: Gaussian-based filters struggle with non-Gaussian distributions, while sequential Monte Carlo methods are computationally intensive and prone to particle degeneracy in high dimensions. Although generative models in machine learning have made significant progress in modeling high-dimensional non-Gaussian distributions, their inefficiency in online updating limits their applicability to filtering problems. To address these challenges, we propose a flow-based Bayesian filter (FBF) that integrates normalizing flows to construct a novel latent linear state-space model with Gaussian filtering distributions. This framework facilitates efficient density estimation and sampling using invertible transformations provided by normalizing flows, and it enables the construction of filters in a data-driven manner, without requiring prior knowledge of system dynamics or observation models. Numerical experiments demonstrate the superior accuracy and efficiency of FBF.
♻ ☆ Learning High-Degree Parities: The Crucial Role of the Initialization
Parities have become a standard benchmark for evaluating learning algorithms. Recent works show that regular neural networks trained by gradient descent can efficiently learn degree $k$ parities on uniform inputs for constant $k$, but fail to do so when $k$ and $d-k$ grow with $d$ (here $d$ is the ambient dimension). However, the case where $k=d-O_d(1)$ (almost-full parities), including the degree $d$ parity (the full parity), has remained unsettled. This paper shows that for gradient descent on regular neural networks, learnability depends on the initial weight distribution. On one hand, the discrete Rademacher initialization enables efficient learning of almost-full parities, while on the other hand, its Gaussian perturbation with large enough constant standard deviation $\sigma$ prevents it. The positive result for almost-full parities is shown to hold up to $\sigma=O(d^{-1})$, pointing to questions about a sharper threshold phenomenon. Unlike statistical query (SQ) learning, where a singleton function class like the full parity is trivially learnable, our negative result applies to a fixed function and relies on an initial gradient alignment measure of potential broader relevance to neural networks learning.
♻ ☆ Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator ICLR 2025
Dataset distillation has emerged as a technique aiming to condense informative features from large, natural datasets into a compact and synthetic form. While recent advancements have refined this technique, its performance is bottlenecked by the prevailing class-specific synthesis paradigm. Under this paradigm, synthetic data is optimized exclusively for a pre-assigned one-hot label, creating an implicit class barrier in feature condensation. This leads to inefficient utilization of the distillation budget and oversight of inter-class feature distributions, which ultimately limits the effectiveness and efficiency, as demonstrated in our analysis. To overcome these constraints, this paper presents the Inter-class Feature Compensator (INFER), an innovative distillation approach that transcends the class-specific data-label framework widely utilized in current dataset distillation methods. Specifically, INFER leverages a Universal Feature Compensator (UFC) to enhance feature integration across classes, enabling the generation of multiple additional synthetic instances from a single UFC input. This significantly improves the efficiency of the distillation budget. Moreover, INFER enriches inter-class interactions during the distillation, thereby enhancing the effectiveness and generalizability of the distilled data. By allowing for the linear interpolation of labels similar to those in the original dataset, INFER meticulously optimizes the synthetic data and dramatically reduces the size of soft labels in the synthetic dataset to almost zero, establishing a new benchmark for efficiency and effectiveness in dataset distillation. In practice, INFER demonstrates state-of-the-art performance across benchmark datasets. For instance, in the ipc = 50 setting on ImageNet-1k with the same compression level, it outperforms SRe2L by 34.5% using ResNet18.
comment: Accepted to ICLR 2025
♻ ☆ DP-LDMs: Differentially Private Latent Diffusion Models
Diffusion models (DMs) are one of the most widely used generative models for producing high quality images. However, a flurry of recent papers points out that DMs are least private forms of image generators, by extracting a significant number of near-identical replicas of training images from DMs. Existing privacy-enhancing techniques for DMs, unfortunately, do not provide a good privacy-utility tradeoff. In this paper, we aim to improve the current state of DMs with differential privacy (DP) by adopting the $\textit{Latent}$ Diffusion Models (LDMs). LDMs are equipped with powerful pre-trained autoencoders that map the high-dimensional pixels into lower-dimensional latent representations, in which DMs are trained, yielding a more efficient and fast training of DMs. Rather than fine-tuning the entire LDMs, we fine-tune only the $\textit{attention}$ modules of LDMs with DP-SGD, reducing the number of trainable parameters by roughly $90\%$ and achieving a better privacy-accuracy trade-off. Our approach allows us to generate realistic, high-dimensional images (256x256) conditioned on text prompts with DP guarantees, which, to the best of our knowledge, has not been attempted before. Our approach provides a promising direction for training more powerful, yet training-efficient differentially private DMs, producing high-quality DP images. Our code is available at https://anonymous.4open.science/r/DP-LDM-4525.
♻ ☆ Improved Performances and Motivation in Intelligent Tutoring Systems: Combining Machine Learning and Learner Choice
Large class sizes challenge personalized learning in schools, prompting the use of educational technologies such as intelligent tutoring systems. To address this, we present an AI-driven personalization system, called ZPDES, based on the Learning Progress Hypothesis - modeling curiosity-driven learning - and multi-armed bandit techniques. It sequences exercises that maximize learning progress for each student. While previous studies demonstrated its efficacy in enhancing learning compared to hand-made curricula, its impact on student motivation remained unexplored. Furthermore, ZPDES previously lacked features allowing student choice, a limitation in agency that conflicts with its foundation on models of curiosity-driven learning. This study investigates how integrating choice, as a gamification element unrelated to exercise difficulty, affects both learning outcomes and motivation. We conducted an extensive field study (265 7-8 years old children, RCT design), comparing ZPDES with and without choice against a hand-designed curriculum. Results show that ZPDES improves both learning performance and the learning experience. Moreover adding choice to ZPDES enhances intrinsic motivation and further strengthens its learning benefits. In contrast, incorporating choice into a fixed, linear curriculum negatively impacts learning outcomes. These findings highlight that the intrinsic motivation elicited by choice (gamification) is beneficial only when paired with an adaptive personalized learning system. This insight is critical as gamified features become increasingly prevalent in educational technologies.
♻ ☆ Regularization-based Framework for Quantization-, Fault- and Variability-Aware Training
Efficient inference is critical for deploying deep learning models on edge AI devices. Low-bit quantization (e.g., 3- and 4-bit) with fixed-point arithmetic improves efficiency, while low-power memory technologies like analog nonvolatile memory enable further gains. However, these methods introduce non-ideal hardware behavior, including bit faults and device-to-device variability. We propose a regularization-based quantization-aware training (QAT) framework that supports fixed, learnable step-size, and learnable non-uniform quantization, achieving competitive results on CIFAR-10 and ImageNet. Our method also extends to Spiking Neural Networks (SNNs), demonstrating strong performance on 4-bit networks on CIFAR10-DVS and N-Caltech 101. Beyond quantization, our framework enables fault and variability-aware fine-tuning, mitigating stuck-at faults (fixed weight bits) and device resistance variability. Compared to prior fault-aware training, our approach significantly improves performance recovery under upto 20% bit-fault rate and 40% device-to-device variability. Our results establish a generalizable framework for quantization and robustness-aware training, enhancing efficiency and reliability in low-power, non-ideal hardware.
comment: AB and RS contributed equally to this work. A version of this paper accepted at MLNCP @ NeuRIPS '24
♻ ☆ Grams: Gradient Descent with Adaptive Momentum Scaling SC
We introduce $\mathbf{G}$radient Descent with $\mathbf{A}$daptive $\mathbf{M}$omentum $\mathbf{S}$caling ($\mathbf{Grams}$), a novel optimization algorithm that decouples the direction and magnitude of parameter updates in deep learning. Unlike traditional optimizers that directly integrate momentum into updates, Grams separates the update direction, derived from current gradients, from momentum, which is used solely for adaptive magnitude scaling. This approach enables Grams to achieve improved loss descent compared to state-of-the-art cautious and momentum-based optimizers. We theoretically demonstrate that Grams descents faster than other state-of-the-art optimizers and establish a global convergence guarantee for Grams. We also validate its effectiveness through extensive empirical evaluations. The results demonstrate Grams' superior performance, including faster convergence and better generalization, compared to widely-used optimizers such as Adam, Lion, and their cautious variants. Our results highlight Grams' potential as a transformative approach for efficiently training and fine-tuning large language models. Code is available at https://github.com/Gunale0926/Grams.
comment: SCOPE Workshop @ ICLR 2025
♻ ☆ Training a Generally Curious Agent
Efficient exploration is essential for intelligent systems interacting with their environment, but existing language models often fall short in scenarios that require strategic information gathering. In this paper, we present PAPRIKA, a fine-tuning approach that enables language models to develop general decision-making capabilities that are not confined to particular environments. By training on synthetic interaction data from different tasks that require diverse strategies, PAPRIKA teaches models to explore and adapt their behavior on a new task based on environment feedback in-context without more gradient updates. Experimental results show that models fine-tuned with PAPRIKA can effectively transfer their learned decision-making capabilities to entirely unseen tasks without additional training. Unlike traditional training, our approach's primary bottleneck lies in sampling useful interaction data instead of model updates. To improve sample efficiency, we propose a curriculum learning strategy that prioritizes sampling trajectories from tasks with high learning potential. These results suggest a promising path towards AI systems that can autonomously solve novel sequential decision-making problems that require interactions with the external world.
comment: Project Website: https://paprika-llm.github.io
♻ ☆ Affordance-Guided Reinforcement Learning via Visual Prompting
Robots equipped with reinforcement learning (RL) have the potential to learn a wide range of skills solely from a reward signal. However, obtaining a robust and dense reward signal for general manipulation tasks remains a challenge. Existing learning-based approaches require significant data, such as human demonstrations of success and failure, to learn task-specific reward functions. Recently, there is also a growing adoption of large multi-modal foundation models for robotics that can perform visual reasoning in physical contexts and generate coarse robot motions for manipulation tasks. Motivated by this range of capability, in this work, we present Keypoint-based Affordance Guidance for Improvements (KAGI), a method leveraging rewards shaped by vision-language models (VLMs) for autonomous RL. State-of-the-art VLMs have demonstrated impressive reasoning about affordances through keypoints in zero-shot, and we use these to define dense rewards that guide autonomous robotic learning. On real-world manipulation tasks specified by natural language descriptions, KAGI improves the sample efficiency of autonomous RL and enables successful task completion in 30K online fine-tuning steps. Additionally, we demonstrate the robustness of KAGI to reductions in the number of in-domain demonstrations used for pre-training, reaching similar performance in 45K online fine-tuning steps. Project website: https://sites.google.com/view/affordance-guided-rl
comment: 8 pages, 6 figures. Robotics: Science and Systems (RSS) 2024, Task Specification for General-Purpose Intelligent Robots & Lifelong Robot Learning Workshops
♻ ☆ $μ^2$-SGD: Stable Stochastic Optimization via a Double Momentum Mechanism
We consider stochastic convex optimization problems where the objective is an expectation over smooth functions. For this setting we suggest a novel gradient estimate that combines two recent mechanism that are related to notion of momentum. Then, we design an SGD-style algorithm as well as an accelerated version that make use of this new estimator, and demonstrate the robustness of these new approaches to the choice of the learning rate. Concretely, we show that these approaches obtain the optimal convergence rates for both noiseless and noisy case with the same choice of fixed learning rate. Moreover, for the noisy case we show that these approaches achieve the same optimal bound for a very wide range of learning rates.
♻ ☆ CarPlanner: Consistent Auto-regressive Trajectory Planning for Large-scale Reinforcement Learning in Autonomous Driving CVPR 2025
Trajectory planning is vital for autonomous driving, ensuring safe and efficient navigation in complex environments. While recent learning-based methods, particularly reinforcement learning (RL), have shown promise in specific scenarios, RL planners struggle with training inefficiencies and managing large-scale, real-world driving scenarios. In this paper, we introduce \textbf{CarPlanner}, a \textbf{C}onsistent \textbf{a}uto-\textbf{r}egressive \textbf{Planner} that uses RL to generate multi-modal trajectories. The auto-regressive structure enables efficient large-scale RL training, while the incorporation of consistency ensures stable policy learning by maintaining coherent temporal consistency across time steps. Moreover, CarPlanner employs a generation-selection framework with an expert-guided reward function and an invariant-view module, simplifying RL training and enhancing policy performance. Extensive analysis demonstrates that our proposed RL framework effectively addresses the challenges of training efficiency and performance enhancement, positioning CarPlanner as a promising solution for trajectory planning in autonomous driving. To the best of our knowledge, we are the first to demonstrate that the RL-based planner can surpass both IL- and rule-based state-of-the-arts (SOTAs) on the challenging large-scale real-world dataset nuPlan. Our proposed CarPlanner surpasses RL-, IL-, and rule-based SOTA approaches within this demanding dataset.
comment: CVPR 2025
♻ ☆ CPT-Boosted Wav2vec2.0: Towards Noise Robust Speech Recognition for Classroom Environments
Creating Automatic Speech Recognition (ASR) systems that are robust and resilient to classroom conditions is paramount to the development of AI tools to aid teachers and students. In this work, we study the efficacy of continued pretraining (CPT) in adapting Wav2vec2.0 to the classroom domain. We show that CPT is a powerful tool in that regard and reduces the Word Error Rate (WER) of Wav2vec2.0-based models by upwards of 10%. More specifically, CPT improves the model's robustness to different noises, microphones and classroom conditions.
comment: arXiv admin note: substantial text overlap with arXiv:2405.13018
♻ ☆ Online Planning for Multi-UAV Pursuit-Evasion in Unknown Environments Using Deep Reinforcement Learning
Multi-UAV pursuit-evasion, where pursuers aim to capture evaders, poses a key challenge for UAV swarm intelligence. Multi-agent reinforcement learning (MARL) has demonstrated potential in modeling cooperative behaviors, but most RL-based approaches remain constrained to simplified simulations with limited dynamics or fixed scenarios. Previous attempts to deploy RL policy to real-world pursuit-evasion are largely restricted to two-dimensional scenarios, such as ground vehicles or UAVs at fixed altitudes. In this paper, we address multi-UAV pursuit-evasion by considering UAV dynamics and physical constraints. We introduce an evader prediction-enhanced network to tackle partial observability in cooperative strategy learning. Additionally, we propose an adaptive environment generator within MARL training, enabling higher exploration efficiency and better policy generalization across diverse scenarios. Simulations show our method significantly outperforms all baselines in challenging scenarios, generalizing to unseen scenarios with a 100% capture rate. Finally, we derive a feasible policy via a two-stage reward refinement and deploy the policy on real quadrotors in a zero-shot manner. To our knowledge, this is the first work to derive and deploy an RL-based policy using collective thrust and body rates control commands for multi-UAV pursuit-evasion in unknown environments. The open-source code and videos are available at https://sites.google.com/view/pursuit-evasion-rl.
♻ ☆ LoBAM: LoRA-Based Backdoor Attack on Model Merging
Model merging is an emerging technique that integrates multiple models fine-tuned on different tasks to create a versatile model that excels in multiple domains. This scheme, in the meantime, may open up backdoor attack opportunities where one single malicious model can jeopardize the integrity of the merged model. Existing works try to demonstrate the risk of such attacks by assuming substantial computational resources, focusing on cases where the attacker can fully fine-tune the pre-trained model. Such an assumption, however, may not be feasible given the increasing size of machine learning models. In practice where resources are limited and the attacker can only employ techniques like Low-Rank Adaptation (LoRA) to produce the malicious model, it remains unclear whether the attack can still work and pose threats. In this work, we first identify that the attack efficacy is significantly diminished when using LoRA for fine-tuning. Then, we propose LoBAM, a method that yields high attack success rate with minimal training resources. The key idea of LoBAM is to amplify the malicious weights in an intelligent way that effectively enhances the attack efficacy. We demonstrate that our design can lead to improved attack success rate through extensive empirical experiments across various model merging scenarios. Moreover, we show that our method is highly stealthy and is difficult to detect and defend against.
Computation and Language 151
☆ Wikipedia in the Era of LLMs: Evolution and Risks
In this paper, we present a thorough analysis of the impact of Large Language Models (LLMs) on Wikipedia, examining the evolution of Wikipedia through existing data and using simulations to explore potential risks. We begin by analyzing page views and article content to study Wikipedia's recent changes and assess the impact of LLMs. Subsequently, we evaluate how LLMs affect various Natural Language Processing (NLP) tasks related to Wikipedia, including machine translation and retrieval-augmented generation (RAG). Our findings and simulation results reveal that Wikipedia articles have been influenced by LLMs, with an impact of approximately 1%-2% in certain categories. If the machine translation benchmark based on Wikipedia is influenced by LLMs, the scores of the models may become inflated, and the comparative results among models might shift as well. Moreover, the effectiveness of RAG might decrease if the knowledge base becomes polluted by LLM-generated content. While LLMs have not yet fully changed Wikipedia's language and knowledge structures, we believe that our empirical findings signal the need for careful consideration of potential future risks.
comment: We release all the experimental dataset and source code at: https://github.com/HSM316/LLM_Wikipedia
☆ Language Models can Self-Improve at State-Value Estimation for Better Search
Collecting ground truth task completion rewards or human demonstrations for multi-step reasoning tasks is often cost-prohibitive and time-consuming, especially in interactive domains like web tasks. To address this bottleneck, we present self-taught lookahead, a self-supervised method that leverages state-transition dynamics to train a value model capable of effectively guiding language model-controlled search. We find that moderately sized (8 billion parameters) open-weight value models improved with self-taught lookahead can match the performance of using a frontier LLM such as gpt-4o as the value model. Furthermore, we find that self-taught lookahead improves performance by 20% while reducing costs 37x compared to previous LLM-based tree search, without relying on ground truth rewards.
☆ The First Few Tokens Are All You Need: An Efficient and Effective Unsupervised Prefix Fine-Tuning Method for Reasoning Models
Improving the reasoning capabilities of large language models (LLMs) typically requires supervised fine-tuning with labeled data or computationally expensive sampling. We introduce Unsupervised Prefix Fine-Tuning (UPFT), which leverages the observation of Prefix Self-Consistency -- the shared initial reasoning steps across diverse solution trajectories -- to enhance LLM reasoning efficiency. By training exclusively on the initial prefix substrings (as few as 8 tokens), UPFT removes the need for labeled data or exhaustive sampling. Experiments on reasoning benchmarks show that UPFT matches the performance of supervised methods such as Rejection Sampling Fine-Tuning, while reducing training time by 75% and sampling cost by 99%. Further analysis reveals that errors tend to appear in later stages of the reasoning process and that prefix-based training preserves the model's structural knowledge. This work demonstrates how minimal unsupervised fine-tuning can unlock substantial reasoning gains in LLMs, offering a scalable and resource-efficient alternative to conventional approaches.
☆ FairSense-AI: Responsible AI Meets Sustainability
In this paper, we introduce FairSense-AI: a multimodal framework designed to detect and mitigate bias in both text and images. By leveraging Large Language Models (LLMs) and Vision-Language Models (VLMs), FairSense-AI uncovers subtle forms of prejudice or stereotyping that can appear in content, providing users with bias scores, explanatory highlights, and automated recommendations for fairness enhancements. In addition, FairSense-AI integrates an AI risk assessment component that aligns with frameworks like the MIT AI Risk Repository and NIST AI Risk Management Framework, enabling structured identification of ethical and safety concerns. The platform is optimized for energy efficiency via techniques such as model pruning and mixed-precision computation, thereby reducing its environmental footprint. Through a series of case studies and applications, we demonstrate how FairSense-AI promotes responsible AI use by addressing both the social dimension of fairness and the pressing need for sustainability in large-scale AI deployments. https://vectorinstitute.github.io/FairSense-AI, https://pypi.org/project/fair-sense-ai/
☆ Calibrating LLM Confidence with Semantic Steering: A Multi-Prompt Aggregation Framework
Large Language Models (LLMs) often exhibit misaligned confidence scores, usually overestimating the reliability of their predictions. While verbalized confidence in Large Language Models (LLMs) has gained attention, prior work remains divided on whether confidence scores can be systematically steered through prompting. Recent studies even argue that such prompt-induced confidence shifts are negligible, suggesting LLMs' confidence calibration is rigid to linguistic interventions. Contrary to these claims, we first rigorously confirm the existence of directional confidence shifts by probing three models (including GPT3.5, LLAMA3-70b, GPT4) across 7 benchmarks, demonstrating that explicit instructions can inflate or deflate confidence scores in a regulated manner. Based on this observation, we propose a novel framework containing three components: confidence steering, steered confidence aggregation and steered answers selection, named SteeringConf. Our method, SteeringConf, leverages a confidence manipulation mechanism to steer the confidence scores of LLMs in several desired directions, followed by a summarization module that aggregates the steered confidence scores to produce a final prediction. We evaluate our method on 7 benchmarks and it consistently outperforms the baselines in terms of calibration metrics in task of confidence calibration and failure detection.
☆ (How) Do Language Models Track State?
Transformer language models (LMs) exhibit behaviors -- from storytelling to code generation -- that appear to require tracking the unobserved state of an evolving world. How do they do so? We study state tracking in LMs trained or fine-tuned to compose permutations (i.e., to compute the order of a set of objects after a sequence of swaps). Despite the simple algebraic structure of this problem, many other tasks (e.g., simulation of finite automata and evaluation of boolean expressions) can be reduced to permutation composition, making it a natural model for state tracking in general. We show that LMs consistently learn one of two state tracking mechanisms for this task. The first closely resembles the "associative scan" construction used in recent theoretical work by Liu et al. (2023) and Merrill et al. (2024). The second uses an easy-to-compute feature (permutation parity) to partially prune the space of outputs, then refines this with an associative scan. The two mechanisms exhibit markedly different robustness properties, and we show how to steer LMs toward one or the other with intermediate training tasks that encourage or suppress the heuristics. Our results demonstrate that transformer LMs, whether pretrained or fine-tuned, can learn to implement efficient and interpretable state tracking mechanisms, and the emergence of these mechanisms can be predicted and controlled.
comment: 21 pages, 17 figures, 1 table
☆ Shakespearean Sparks: The Dance of Hallucination and Creativity in LLMs' Decoding Layers
Large language models (LLMs) are known to hallucinate, a phenomenon often linked to creativity. While previous research has primarily explored this connection through theoretical or qualitative lenses, our work takes a quantitative approach to systematically examine the relationship between hallucination and creativity in LLMs. Given the complex nature of creativity, we propose a narrow definition tailored to LLMs and introduce an evaluation framework, HCL, which quantifies Hallucination and Creativity across different Layers of LLMs during decoding. Our empirical analysis reveals a tradeoff between hallucination and creativity that is consistent across layer depth, model type, and model size. Notably, across different model architectures, we identify a specific layer at each model size that optimally balances this tradeoff. Additionally, the optimal layer tends to appear in the early layers of larger models, and the confidence of the model is also significantly higher at this layer. These findings provide a quantitative perspective that offers new insights into the interplay between LLM creativity and hallucination. The code and data for our experiments are available at https://github.com/ZicongHe2002/HCL-Spark.
☆ Mask-DPO: Generalizable Fine-grained Factuality Alignment of LLMs ICLR 2025
Large language models (LLMs) exhibit hallucinations (i.e., unfaithful or nonsensical information) when serving as AI assistants in various domains. Since hallucinations always come with truthful content in the LLM responses, previous factuality alignment methods that conduct response-level preference learning inevitably introduced noises during training. Therefore, this paper proposes a fine-grained factuality alignment method based on Direct Preference Optimization (DPO), called Mask-DPO. Incorporating sentence-level factuality as mask signals, Mask-DPO only learns from factually correct sentences in the preferred samples and prevents the penalty on factual contents in the not preferred samples, which resolves the ambiguity in the preference learning. Extensive experimental results demonstrate that Mask-DPO can significantly improve the factuality of LLMs responses to questions from both in-domain and out-of-domain datasets, although these questions and their corresponding topics are unseen during training. Only trained on the ANAH train set, the score of Llama3.1-8B-Instruct on the ANAH test set is improved from 49.19% to 77.53%, even surpassing the score of Llama3.1-70B-Instruct (53.44%), while its FactScore on the out-of-domain Biography dataset is also improved from 30.29% to 39.39%. We further study the generalization property of Mask-DPO using different training sample scaling strategies and find that scaling the number of topics in the dataset is more effective than the number of questions. We provide a hypothesis of what factual alignment is doing with LLMs, on the implication of this phenomenon, and conduct proof-of-concept experiments to verify it. We hope the method and the findings pave the way for future research on scaling factuality alignment.
comment: Accepted by ICLR 2025. Code is available at https://github.com/open-compass/ANAH
☆ AlignDistil: Token-Level Language Model Alignment as Adaptive Policy Distillation
In modern large language models (LLMs), LLM alignment is of crucial importance and is typically achieved through methods such as reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO). However, in most existing methods for LLM alignment, all tokens in the response are optimized using a sparse, response-level reward or preference annotation. The ignorance of token-level rewards may erroneously punish high-quality tokens or encourage low-quality tokens, resulting in suboptimal performance and slow convergence speed. To address this issue, we propose AlignDistil, an RLHF-equivalent distillation method for token-level reward optimization. Specifically, we introduce the reward learned by DPO into the RLHF objective and theoretically prove the equivalence between this objective and a token-level distillation process, where the teacher distribution linearly combines the logits from the DPO model and a reference model. On this basis, we further bridge the accuracy gap between the reward from the DPO model and the pure reward model, by building a contrastive DPO reward with a normal and a reverse DPO model. Moreover, to avoid under- and over-optimization on different tokens, we design a token adaptive logit extrapolation mechanism to construct an appropriate teacher distribution for each token. Experimental results demonstrate the superiority of our AlignDistil over existing methods and showcase fast convergence due to its token-level distributional reward optimization.
comment: 15 pages, 2 figures
☆ Q-Filters: Leveraging QK Geometry for Efficient KV Cache Compression
Autoregressive language models rely on a Key-Value (KV) Cache, which avoids re-computing past hidden states during generation, making it faster. As model sizes and context lengths grow, the KV Cache becomes a significant memory bottleneck, which calls for compression methods that limit its size during generation. In this paper, we discover surprising properties of Query (Q) and Key (K) vectors that allow us to efficiently approximate attention scores without computing the attention maps. We propose Q-Filters, a training-free KV Cache compression method that filters out less crucial Key-Value pairs based on a single context-agnostic projection. Contrarily to many alternatives, Q-Filters is compatible with FlashAttention, as it does not require direct access to attention weights. Experimental results in long-context settings demonstrate that Q-Filters is competitive with attention-based compression methods such as SnapKV in retrieval tasks while consistently outperforming efficient compression schemes such as Streaming-LLM in generation setups. Notably, Q-Filters achieves a 99% accuracy in the needle-in-a-haystack task with a x32 compression level while reducing the generation perplexity drop by up to 65% in text generation compared to Streaming-LLM.
☆ IterPref: Focal Preference Learning for Code Generation via Iterative Debugging
Preference learning enhances Code LLMs beyond supervised fine-tuning by leveraging relative quality comparisons. Existing methods construct preference pairs from candidates based on test case success, treating the higher pass rate sample as positive and the lower as negative. However, this approach does not pinpoint specific errors in the code, which prevents the model from learning more informative error correction patterns, as aligning failing code as a whole lacks the granularity needed to capture meaningful error-resolution relationships. To address these issues, we propose IterPref, a new preference alignment framework that mimics human iterative debugging to refine Code LLMs. IterPref explicitly locates error regions and aligns the corresponding tokens via a tailored DPO algorithm. To generate informative pairs, we introduce the CodeFlow dataset, where samples are iteratively refined until passing tests, with modifications capturing error corrections. Extensive experiments show that a diverse suite of Code LLMs equipped with IterPref achieves significant performance gains in code generation and improves on challenging tasks like BigCodeBench. In-depth analysis reveals that IterPref yields fewer errors. Our code and data will be made publicaly available.
comment: The code and data will be released soon
☆ Implicit Bias in LLMs: A Survey
Due to the implement of guardrails by developers, Large language models (LLMs) have demonstrated exceptional performance in explicit bias tests. However, bias in LLMs may occur not only explicitly, but also implicitly, much like humans who consciously strive for impartiality yet still harbor implicit bias. The unconscious and automatic nature of implicit bias makes it particularly challenging to study. This paper provides a comprehensive review of the existing literature on implicit bias in LLMs. We begin by introducing key concepts, theories and methods related to implicit bias in psychology, extending them from humans to LLMs. Drawing on the Implicit Association Test (IAT) and other psychological frameworks, we categorize detection methods into three primary approaches: word association, task-oriented text generation and decision-making. We divide our taxonomy of evaluation metrics for implicit bias into two categories: single-value-based metrics and comparison-value-based metrics. We classify datasets into two types: sentences with masked tokens and complete sentences, incorporating datasets from various domains to reflect the broad application of LLMs. Although research on mitigating implicit bias in LLMs is still limited, we summarize existing efforts and offer insights on future challenges. We aim for this work to serve as a clear guide for researchers and inspire innovative ideas to advance exploration in this task.
☆ InSerter: Speech Instruction Following with Unsupervised Interleaved Pre-training
Recent advancements in speech large language models (SpeechLLMs) have attracted considerable attention. Nonetheless, current methods exhibit suboptimal performance in adhering to speech instructions. Notably, the intelligence of models significantly diminishes when processing speech-form input as compared to direct text-form input. Prior work has attempted to mitigate this semantic inconsistency between speech and text representations through techniques such as representation and behavior alignment, which involve the meticulous design of data pairs during the post-training phase. In this paper, we introduce a simple and scalable training method called InSerter, which stands for Interleaved Speech-Text Representation Pre-training. InSerter is designed to pre-train large-scale unsupervised speech-text sequences, where the speech is synthesized from randomly selected segments of an extensive text corpus using text-to-speech conversion. Consequently, the model acquires the ability to generate textual continuations corresponding to the provided speech segments, obviating the need for intensive data design endeavors. To systematically evaluate speech instruction-following capabilities, we introduce SpeechInstructBench, the first comprehensive benchmark specifically designed for speech-oriented instruction-following tasks. Our proposed InSerter achieves SOTA performance in SpeechInstructBench and demonstrates superior or competitive results across diverse speech processing tasks.
☆ From Metaphor to Mechanism: How LLMs Decode Traditional Chinese Medicine Symbolic Language for Modern Clinical Relevance
Metaphorical expressions are abundant in Traditional Chinese Medicine (TCM), conveying complex disease mechanisms and holistic health concepts through culturally rich and often abstract terminology. Bridging these metaphors to anatomically driven Western medical (WM) concepts poses significant challenges for both automated language processing and real-world clinical practice. To address this gap, we propose a novel multi-agent and chain-of-thought (CoT) framework designed to interpret TCM metaphors accurately and map them to WM pathophysiology. Specifically, our approach combines domain-specialized agents (TCM Expert, WM Expert) with a Coordinator Agent, leveraging stepwise chain-of-thought prompts to ensure transparent reasoning and conflict resolution. We detail a methodology for building a metaphor-rich TCM dataset, discuss strategies for effectively integrating multi-agent collaboration and CoT reasoning, and articulate the theoretical underpinnings that guide metaphor interpretation across distinct medical paradigms. We present a comprehensive system design and highlight both the potential benefits and limitations of our approach, while leaving placeholders for future experimental validation. Our work aims to support clinical decision-making, cross-system educational initiatives, and integrated healthcare research, ultimately offering a robust scaffold for reconciling TCM's symbolic language with the mechanistic focus of Western medicine.
☆ BatchGEMBA: Token-Efficient Machine Translation Evaluation with Batched Prompting and Prompt Compression
Recent advancements in Large Language Model (LLM)-based Natural Language Generation evaluation have largely focused on single-example prompting, resulting in significant token overhead and computational inefficiencies. In this work, we introduce BatchGEMBA-MQM, a framework that integrates batched prompting with the GEMBA-MQM metric for machine translation evaluation. Our approach aggregates multiple translation examples into a single prompt, reducing token usage by 2-4 times (depending on the batch size) relative to single-example prompting. Furthermore, we propose a batching-aware prompt compression model that achieves an additional token reduction of 13-15% on average while also showing ability to help mitigate batching-induced quality degradation. Evaluations across several LLMs (GPT-4o, GPT-4o-mini, Mistral Small, Phi4, and CommandR7B) and varying batch sizes reveal that while batching generally negatively affects quality (but sometimes not substantially), prompt compression does not degrade further, and in some cases, recovers quality loss. For instance, GPT-4o retains over 90% of its baseline performance at a batch size of 4 when compression is applied, compared to a 44.6% drop without compression. We plan to release our code and trained models at https://github.com/NL2G/batchgemba to support future research in this domain.
☆ Seeded Poisson Factorization: Leveraging domain knowledge to fit topic models
Topic models are widely used for discovering latent thematic structures in large text corpora, yet traditional unsupervised methods often struggle to align with predefined conceptual domains. This paper introduces Seeded Poisson Factorization (SPF), a novel approach that extends the Poisson Factorization framework by incorporating domain knowledge through seed words. SPF enables a more interpretable and structured topic discovery by modifying the prior distribution of topic-specific term intensities, assigning higher initial rates to predefined seed words. The model is estimated using variational inference with stochastic gradient optimization, ensuring scalability to large datasets. We apply SPF to an Amazon customer feedback dataset, leveraging predefined product categories as guiding structures. Our evaluation demonstrates that SPF achieves superior classification performance compared to alternative guided topic models, particularly in terms of computational efficiency and predictive performance. Furthermore, robustness checks highlight SPF's ability to adaptively balance domain knowledge and data-driven topic discovery, even in cases of imperfect seed word selection. These results establish SPF as a powerful and scalable alternative for integrating expert knowledge into topic modeling, enhancing both interpretability and efficiency in real-world applications.
Large Language Models for Multilingual Previously Fact-Checked Claim Detection
In our era of widespread false information, human fact-checkers often face the challenge of duplicating efforts when verifying claims that may have already been addressed in other countries or languages. As false information transcends linguistic boundaries, the ability to automatically detect previously fact-checked claims across languages has become an increasingly important task. This paper presents the first comprehensive evaluation of large language models (LLMs) for multilingual previously fact-checked claim detection. We assess seven LLMs across 20 languages in both monolingual and cross-lingual settings. Our results show that while LLMs perform well for high-resource languages, they struggle with low-resource languages. Moreover, translating original texts into English proved to be beneficial for low-resource languages. These findings highlight the potential of LLMs for multilingual previously fact-checked claim detection and provide a foundation for further research on this promising application of LLMs.
☆ Evaluating Knowledge Generation and Self-Refinement Strategies for LLM-based Column Type Annotation
Understanding the semantics of columns in relational tables is an important pre-processing step for indexing data lakes in order to provide rich data search. An approach to establishing such understanding is column type annotation (CTA) where the goal is to annotate table columns with terms from a given vocabulary. This paper experimentally compares different knowledge generation and self-refinement strategies for LLM-based column type annotation. The strategies include using LLMs to generate term definitions, error-based refinement of term definitions, self-correction, and fine-tuning using examples and term definitions. We evaluate these strategies along two dimensions: effectiveness measured as F1 performance and efficiency measured in terms of token usage and cost. Our experiments show that the best performing strategy depends on the model/dataset combination. We find that using training data to generate label definitions outperforms using the same data as demonstrations for in-context learning for two out of three datasets using OpenAI models. The experiments further show that using the LLMs to refine label definitions brings an average increase of 3.9% F1 in 10 out of 12 setups compared to the performance of the non-refined definitions. Combining fine-tuned models with self-refined term definitions results in the overall highest performance, outperforming zero-shot prompting fine-tuned models by at least 3% in F1 score. The costs analysis shows that while reaching similar F1 score, self-refinement via prompting is more cost efficient for use cases requiring smaller amounts of tables to be annotated while fine-tuning is more efficient for large amounts of tables.
☆ Multilingualism, Transnationality, and K-pop in the Online #StopAsianHate Movement
The #StopAsianHate (SAH) movement is a broad social movement against violence targeting Asians and Asian Americans, beginning in 2021 in response to racial discrimination related to COVID-19 and sparking worldwide conversation about anti-Asian hate. However, research on the online SAH movement has focused on English-speaking participants so the spread of the movement outside of the United States is largely unknown. In addition, there have been no long-term studies of SAH so the extent to which it has been successfully sustained over time is not well understood. We present an analysis of 6.5 million "#StopAsianHate" tweets from 2.2 million users all over the globe and spanning 60 different languages, constituting the first study of the non-English and transnational component of the online SAH movement. Using a combination of topic modeling, user modeling, and hand annotation, we identify and characterize the dominant discussions and users participating in the movement and draw comparisons of English versus non-English topics and users. We discover clear differences in events driving topics, where spikes in English tweets are driven by violent crimes in the US but spikes in non-English tweets are driven by transnational incidents of anti-Asian sentiment towards symbolic representatives of Asian nations. We also find that global K-pop fans were quick to adopt the SAH movement and, in fact, sustained it for longer than any other user group. Our work contributes to understanding the transnationality and evolution of the SAH movement, and more generally to exploring upward scale shift and public attention in large-scale multilingual online activism.
comment: WebSci'25
☆ MPO: Boosting LLM Agents with Meta Plan Optimization
Recent advancements in large language models (LLMs) have enabled LLM-based agents to successfully tackle interactive planning tasks. However, despite their successes, existing approaches often suffer from planning hallucinations and require retraining for each new agent. To address these challenges, we propose the Meta Plan Optimization (MPO) framework, which enhances agent planning capabilities by directly incorporating explicit guidance. Unlike previous methods that rely on complex knowledge, which either require significant human effort or lack quality assurance, MPO leverages high-level general guidance through meta plans to assist agent planning and enables continuous optimization of the meta plans based on feedback from the agent's task execution. Our experiments conducted on two representative tasks demonstrate that MPO significantly outperforms existing baselines. Moreover, our analysis indicates that MPO provides a plug-and-play solution that enhances both task completion efficiency and generalization capabilities in previous unseen scenarios.
☆ Are some books better than others?
Scholars, awards committees, and laypeople frequently discuss the merit of written works. Literary professionals and journalists differ in how much perspectivism they concede in their book reviews. Here, we quantify how strongly book reviews are determined by the actual book contents vs. idiosyncratic reader tendencies. In our analysis of 624,320 numerical and textual book reviews, we find that the contents of professionally published books are not predictive of a random reader's reading enjoyment. Online reviews of popular fiction and non-fiction books carry up to ten times more information about the reviewer than about the book. For books of a preferred genre, readers might be less likely to give low ratings, but still struggle to converge in their relative assessments. We find that book evaluations generalize more across experienced review writers than casual readers. When discussing specific issues with a book, one review text had poor predictability of issues brought up in another review of the same book. We conclude that extreme perspectivism is a justifiable position when researching literary quality, bestowing literary awards, and designing recommendation systems.
☆ Multidimensional Consistency Improves Reasoning in Language Models
While Large language models (LLMs) have proved able to address some complex reasoning tasks, we also know that they are highly sensitive to input variation, which can lead to different solution paths and final answers. Answer consistency across input variations can thus be taken as a sign of stronger confidence. Leveraging this insight, we introduce a framework, {\em Multidimensional Reasoning Consistency} where, focusing on math problems, models are systematically pushed to diversify solution paths towards a final answer, thereby testing them for answer consistency across multiple input variations. We induce variations in (i) order of shots in prompt, (ii) problem phrasing, and (iii) languages used. Extensive experiments on a large range of open-source state-of-the-art LLMs of various sizes show that reasoning consistency differs by variation dimension, and that by aggregating consistency across dimensions, our framework consistently enhances mathematical reasoning performance on both monolingual dataset GSM8K and multilingual dataset MGSM, especially for smaller models.
☆ LoRA-Null: Low-Rank Adaptation via Null Space for Large Language Models
Low-Rank Adaptation (LoRA) is the leading parameter-efficient fine-tuning method for Large Language Models (LLMs). However, the fine-tuned LLMs encounter the issue of catastrophic forgetting of the pre-trained world knowledge. To address this issue, inspired by theoretical insights of null space, we propose LoRA-Null, i.e., Low-Rank Adaptation via null space, which builds adapters initialized from the null space of the pre-trained knowledge activation. Concretely, we randomly collect a few data samples and capture their activations after passing through the LLM layer. We perform Singular Value Decomposition on the input activations to obtain their null space. We use the projection of the pre-trained weights onto the null space as the initialization for adapters. Experimental results demonstrate that this initialization approach can effectively preserve the original pre-trained world knowledge of the LLMs during fine-tuning. Additionally, if we freeze the values of the down-projection matrices during fine-tuning, it achieves even better preservation of the pre-trained world knowledge. LoRA-Null effectively preserves pre-trained world knowledge while maintaining strong fine-tuning performance, as validated by extensive experiments on LLaMA series (LLaMA2, LLaMA3, LLaMA3.1, and LLaMA3.2) across Code, Math, and Instruction Following tasks. We also provide a theoretical guarantee for the capacity of LoRA-Null to retain pre-trained knowledge. Code is in https://github.com/HungerPWAY/LoRA-Null.
☆ Adapting Decoder-Based Language Models for Diverse Encoder Downstream Tasks
Decoder-based transformers, while revolutionizing language modeling and scaling to immense sizes, have not completely overtaken encoder-heavy architectures in natural language processing. Specifically, encoder-only models remain dominant in tasks like classification, regression, and ranking. This is primarily due to the inherent structure of decoder-based models, which limits their direct applicability to these tasks. In this paper, we introduce Gemma Encoder, adapting the powerful Gemma decoder model to an encoder architecture, thereby unlocking its potential for a wider range of non-generative applications. To optimize the adaptation from decoder to encoder, we systematically analyze various pooling strategies, attention mechanisms, and hyperparameters (e.g., dropout rate). Furthermore, we benchmark Gemma Encoder against established approaches on the GLUE benchmarks, and MS MARCO ranking benchmark, demonstrating its effectiveness and versatility.
☆ The Effectiveness of Large Language Models in Transforming Unstructured Text to Standardized Formats
The exponential growth of unstructured text data presents a fundamental challenge in modern data management and information retrieval. While Large Language Models (LLMs) have shown remarkable capabilities in natural language processing, their potential to transform unstructured text into standardized, structured formats remains largely unexplored - a capability that could revolutionize data processing workflows across industries. This study breaks new ground by systematically evaluating LLMs' ability to convert unstructured recipe text into the structured Cooklang format. Through comprehensive testing of four models (GPT-4o, GPT-4o-mini, Llama3.1:70b, and Llama3.1:8b), an innovative evaluation approach is introduced that combines traditional metrics (WER, ROUGE-L, TER) with specialized metrics for semantic element identification. Our experiments reveal that GPT-4o with few-shot prompting achieves breakthrough performance (ROUGE-L: 0.9722, WER: 0.0730), demonstrating for the first time that LLMs can reliably transform domain-specific unstructured text into structured formats without extensive training. Although model performance generally scales with size, we uncover surprising potential in smaller models like Llama3.1:8b for optimization through targeted fine-tuning. These findings open new possibilities for automated structured data generation across various domains, from medical records to technical documentation, potentially transforming the way organizations process and utilize unstructured information.
☆ Towards Event Extraction with Massive Types: LLM-based Collaborative Annotation and Partitioning Extraction
Developing a general-purpose extraction system that can extract events with massive types is a long-standing target in Event Extraction (EE). In doing so, the challenge comes from two aspects: 1) The absence of an efficient and effective annotation method. 2) The absence of a powerful extraction method can handle massive types. For the first challenge, we propose a collaborative annotation method based on Large Language Models (LLMs). Through collaboration among multiple LLMs, it first refines annotations of trigger words from distant supervision and then carries out argument annotation. Next, a voting phase consolidates the annotation preferences across different LLMs. Finally, we create the EEMT dataset, the largest EE dataset to date, featuring over 200,000 samples, 3,465 event types, and 6,297 role types. For the second challenge, we propose an LLM-based Partitioning EE method called LLM-PEE. To overcome the limited context length of LLMs, LLM-PEE first recalls candidate event types and then splits them into multiple partitions for LLMs to extract events. The results in the supervised setting show that LLM-PEE outperforms the state-of-the-art methods by 5.4 in event detection and 6.1 in argument extraction. In the zero-shot setting, LLM-PEE achieves up to 12.9 improvement compared to mainstream LLMs, demonstrating its strong generalization capabilities.
comment: Work in progress
☆ ttta: Tools for Temporal Text Analysis
Text data is inherently temporal. The meaning of words and phrases changes over time, and the context in which they are used is constantly evolving. This is not just true for social media data, where the language used is rapidly influenced by current events, memes and trends, but also for journalistic, economic or political text data. Most NLP techniques however consider the corpus at hand to be homogenous in regard to time. This is a simplification that can lead to biased results, as the meaning of words and phrases can change over time. For instance, running a classic Latent Dirichlet Allocation on a corpus that spans several years is not enough to capture changes in the topics over time, but only portraits an "average" topic distribution over the whole time span. Researchers have developed a number of tools for analyzing text data over time. However, these tools are often scattered across different packages and libraries, making it difficult for researchers to use them in a consistent and reproducible way. The ttta package is supposed to serve as a collection of tools for analyzing text data over time.
comment: 4 pages, 2 figures
☆ Rewarding Doubt: A Reinforcement Learning Approach to Confidence Calibration of Large Language Models
A safe and trustworthy use of Large Language Models (LLMs) requires an accurate expression of confidence in their answers. We introduce a novel Reinforcement Learning (RL) approach for LLM calibration that fine-tunes LLMs to elicit calibrated confidence estimations in their answers to factual questions. We model the problem as a betting game where the model predicts a confidence score together with every answer, and design a reward function that penalizes both over and under-confidence. We prove that under our reward design an optimal policy would result in a perfectly calibrated confidence estimation. Our experiments demonstrate significantly improved confidence calibration and generalization to new tasks without re-training, indicating that our approach teaches a general confidence awareness. This approach enables the training of inherently calibrated LLMs.
☆ OkraLong: A Flexible Retrieval-Augmented Framework for Long-Text Query Processing
Large Language Models (LLMs) encounter challenges in efficiently processing long-text queries, as seen in applications like enterprise document analysis and financial report comprehension. While conventional solutions employ long-context processing or Retrieval-Augmented Generation (RAG), they suffer from prohibitive input expenses or incomplete information. Recent advancements adopt context compression and dynamic retrieval loops, but still sacrifice critical details or incur iterative costs.To address these limitations, we propose OkraLong, a novel framework that flexibly optimizes the entire processing workflow. Unlike prior static or coarse-grained adaptive strategies, OkraLong adopts fine-grained orchestration through three synergistic components: analyzer, organizer and executor. The analyzer characterizes the task states, which guide the organizer in dynamically scheduling the workflow. The executor carries out the execution and generates the final answer. Experimental results demonstrate that OkraLong not only enhances answer accuracy but also achieves cost-effectiveness across a variety of datasets.
☆ MciteBench: A Benchmark for Multimodal Citation Text Generation in MLLMs
Multimodal Large Language Models (MLLMs) have advanced in integrating diverse modalities but frequently suffer from hallucination. A promising solution to mitigate this issue is to generate text with citations, providing a transparent chain for verification. However, existing work primarily focuses on generating citations for text-only content, overlooking the challenges and opportunities of multimodal contexts. To address this gap, we introduce MCiteBench, the first benchmark designed to evaluate and analyze the multimodal citation text generation ability of MLLMs. Our benchmark comprises data derived from academic papers and review-rebuttal interactions, featuring diverse information sources and multimodal content. We comprehensively evaluate models from multiple dimensions, including citation quality, source reliability, and answer accuracy. Through extensive experiments, we observe that MLLMs struggle with multimodal citation text generation. We also conduct deep analyses of models' performance, revealing that the bottleneck lies in attributing the correct sources rather than understanding the multimodal content.
☆ Use Me Wisely: AI-Driven Assessment for LLM Prompting Skills Development
The use of large language model (LLM)-powered chatbots, such as ChatGPT, has become popular across various domains, supporting a range of tasks and processes. However, due to the intrinsic complexity of LLMs, effective prompting is more challenging than it may seem. This highlights the need for innovative educational and support strategies that are both widely accessible and seamlessly integrated into task workflows. Yet, LLM prompting is highly task- and domain-dependent, limiting the effectiveness of generic approaches. In this study, we explore whether LLM-based methods can facilitate learning assessments by using ad-hoc guidelines and a minimal number of annotated prompt samples. Our framework transforms these guidelines into features that can be identified within learners' prompts. Using these feature descriptions and annotated examples, we create few-shot learning detectors. We then evaluate different configurations of these detectors, testing three state-of-the-art LLMs and ensembles. We run experiments with cross-validation on a sample of original prompts, as well as tests on prompts collected from task-naive learners. Our results show how LLMs perform on feature detection. Notably, GPT- 4 demonstrates strong performance on most features, while closely related models, such as GPT-3 and GPT-3.5 Turbo (Instruct), show inconsistent behaviors in feature classification. These differences highlight the need for further research into how design choices impact feature selection and prompt detection. Our findings contribute to the fields of generative AI literacy and computer-supported learning assessment, offering valuable insights for both researchers and practitioners.
comment: Preprint accepted for Publication in Educational Technology & Society (ET&S)
☆ Generator-Assistant Stepwise Rollback Framework for Large Language Model Agent
Large language model (LLM) agents typically adopt a step-by-step reasoning framework, in which they interleave the processes of thinking and acting to accomplish the given task. However, this paradigm faces a deep-rooted one-pass issue whereby each generated intermediate thought is plugged into the trajectory regardless of its correctness, which can cause irreversible error propagation. To address the issue, this paper proposes a novel framework called Generator-Assistant Stepwise Rollback (GA-Rollback) to induce better decision-making for LLM agents. Particularly, GA-Rollback utilizes a generator to interact with the environment and an assistant to examine each action produced by the generator, where the assistant triggers a rollback operation upon detection of incorrect actions. Moreover, we introduce two additional strategies tailored for the rollback scenario to further improve its effectiveness. Extensive experiments show that GA-Rollback achieves significant improvements over several strong baselines on three widely used benchmarks. Our analysis further reveals that GA-Rollback can function as a robust plug-and-play module, integrating seamlessly with other methods.
☆ LADM: Long-context Training Data Selection with Attention-based Dependency Measurement for LLMs ACL
Long-context modeling has drawn more and more attention in the area of Large Language Models (LLMs). Continual training with long-context data becomes the de-facto method to equip LLMs with the ability to process long inputs. However, it still remains an open challenge to measure the quality of long-context training data. To address this issue, we propose a Long-context data selection framework with Attention-based Dependency Measurement (LADM), which can efficiently identify high-quality long-context data from a large-scale, multi-domain pre-training corpus. LADM leverages the retrieval capabilities of the attention mechanism to capture contextual dependencies, ensuring a comprehensive quality measurement of long-context data. Experimental results show that our LADM framework significantly boosts the performance of LLMs on multiple long-context tasks with only 1B tokens for continual training.
comment: Submitted to ACL ARR 2024 December
☆ Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed Transformer
Mixture-of-Experts (MoE) enhances model performance while maintaining computational efficiency, making it well-suited for large-scale applications. However, expert in exist MoE paradigm works as an individual, thereby lacking high-quality expert interactions. Moreover, they have not been effectively extended to attention block, which constrains further efficiency improvements. To tackle these issues, we propose Union-of-Experts (UoE), which decomposes transformer into an equitant group of experts, and then implement dynamic routing on input data and experts. Our approach advances MoE design with three key innovations: (1) We conducted equitant expert decomposition on both MLP blocks and attention blocks based on matrix partition in tensor parallelism. (2) We developed two routing paradigms: patch wise data selection and expert selection, to apply routing across different levels. (3) We design the architecture of UoE model, including Selective Multi-Head Attention (SMHA) and Union-of-MLP-Experts (UoME). (4) We develop parallel implementation of UoE's routing and computation operation, and optimize efficiency based on the hardware processing analysis. The experiments demonstrate that the model employed with UoE surpass Full Attention, state-of-art MoEs and efficient transformers in several tasks across image and natural language domains. The source codes are available at https://github.com/YujiaoYang-work/UoE.
comment: 17 pages, 6 figures, 5 tables
☆ It Helps to Take a Second Opinion: Teaching Smaller LLMs to Deliberate Mutually via Selective Rationale Optimisation ICLR 2025
Very large language models (LLMs) such as GPT-4 have shown the ability to handle complex tasks by generating and self-refining step-by-step rationales. Smaller language models (SLMs), typically with < 13B parameters, have been improved by using the data generated from very-large LMs through knowledge distillation. However, various practical constraints such as API costs, copyright, legal and ethical policies restrict using large (often opaque) models to train smaller models for commercial use. Limited success has been achieved at improving the ability of an SLM to explore the space of possible rationales and evaluate them by itself through self-deliberation. To address this, we propose COALITION, a trainable framework that facilitates interaction between two variants of the same SLM and trains them to generate and refine rationales optimized for the end-task. The variants exhibit different behaviors to produce a set of diverse candidate rationales during the generation and refinement steps. The model is then trained via Selective Rationale Optimization (SRO) to prefer generating rationale candidates that maximize the likelihood of producing the ground-truth answer. During inference, COALITION employs a controller to select the suitable variant for generating and refining the rationales. On five different datasets covering mathematical problems, commonsense reasoning, and natural language inference, COALITION outperforms several baselines by up to 5%. Our ablation studies reveal that cross-communication between the two variants performs better than using the single model to self-refine the rationales. We also demonstrate the applicability of COALITION for LMs of varying scales (4B to 14B parameters) and model families (Mistral, Llama, Qwen, Phi). We release the code for this work at https://github.com/Sohanpatnaik106/coalition.
comment: Accepted at ICLR 2025
☆ Measuring What Makes You Unique: Difference-Aware User Modeling for Enhancing LLM Personalization
Personalizing Large Language Models (LLMs) has become a critical step in facilitating their widespread application to enhance individual life experiences. In pursuit of personalization, distilling key preference information from an individual's historical data as instructional preference context to customize LLM generation has emerged as a promising direction. However, these methods face a fundamental limitation by overlooking the inter-user comparative analysis, which is essential for identifying the inter-user differences that truly shape preferences. To address this limitation, we propose Difference-aware Personalization Learning (DPL), a novel approach that emphasizes extracting inter-user differences to enhance LLM personalization. DPL strategically selects representative users for comparison and establishes a structured standard to extract meaningful, task-relevant differences for customizing LLM generation. Extensive experiments on real-world datasets demonstrate that DPL significantly enhances LLM personalization. We release our code at https://github.com/SnowCharmQ/DPL.
☆ BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modelling
Time-series Generation (TSG) is a prominent research area with broad applications in simulations, data augmentation, and counterfactual analysis. While existing methods have shown promise in unconditional single-domain TSG, real-world applications demand for cross-domain approaches capable of controlled generation tailored to domain-specific constraints and instance-level requirements. In this paper, we argue that text can provide semantic insights, domain information and instance-specific temporal patterns, to guide and improve TSG. We introduce ``Text-Controlled TSG'', a task focused on generating realistic time series by incorporating textual descriptions. To address data scarcity in this setting, we propose a novel LLM-based Multi-Agent framework that synthesizes diverse, realistic text-to-TS datasets. Furthermore, we introduce BRIDGE, a hybrid text-controlled TSG framework that integrates semantic prototypes with text description for supporting domain-level guidance. This approach achieves state-of-the-art generation fidelity on 11 of 12 datasets, and improves controllability by 12.52% on MSE and 6.34% MAE compared to no text input generation, highlighting its potential for generating tailored time-series data.
comment: Preprint. Work in progress
☆ AILS-NTUA at SemEval-2025 Task 4: Parameter-Efficient Unlearning for Large Language Models using Data Chunking
The Unlearning Sensitive Content from Large Language Models task aims to remove targeted datapoints from trained models while minimally affecting their general knowledge. In our work, we leverage parameter-efficient, gradient-based unlearning using low-rank (LoRA) adaptation and layer-focused fine-tuning. To further enhance unlearning effectiveness, we employ data chunking, splitting forget data into disjoint partitions and merging them with cyclically sampled retain samples at a pre-defined ratio. Our task-agnostic method achieves an outstanding forget-retain balance, ranking first on leaderboards and significantly outperforming baselines and competing systems.
☆ AILS-NTUA at SemEval-2025 Task 3: Leveraging Large Language Models and Translation Strategies for Multilingual Hallucination Detection
Multilingual hallucination detection stands as an underexplored challenge, which the Mu-SHROOM shared task seeks to address. In this work, we propose an efficient, training-free LLM prompting strategy that enhances detection by translating multilingual text spans into English. Our approach achieves competitive rankings across multiple languages, securing two first positions in low-resource languages. The consistency of our results highlights the effectiveness of our translation strategy for hallucination detection, demonstrating its applicability regardless of the source language.
☆ Hierarchical Re-ranker Retriever (HRR)
Retrieving the right level of context for a given query is a perennial challenge in information retrieval - too large a chunk dilutes semantic specificity, while chunks that are too small lack broader context. This paper introduces the Hierarchical Re-ranker Retriever (HRR), a framework designed to achieve both fine-grained and high-level context retrieval for large language model (LLM) applications. In HRR, documents are split into sentence-level and intermediate-level (512 tokens) chunks to maximize vector-search quality for both short and broad queries. We then employ a reranker that operates on these 512-token chunks, ensuring an optimal balance neither too coarse nor too fine for robust relevance scoring. Finally, top-ranked intermediate chunks are mapped to parent chunks (2048 tokens) to provide an LLM with sufficiently large context.
comment: 14 pages
☆ An Efficient and Precise Training Data Construction Framework for Process-supervised Reward Model in Mathematical Reasoning
Enhancing the mathematical reasoning capabilities of Large Language Models (LLMs) is of great scientific and practical significance. Researchers typically employ process-supervised reward models (PRMs) to guide the reasoning process, effectively improving the models' reasoning abilities. However, existing methods for constructing process supervision training data, such as manual annotation and per-step Monte Carlo estimation, are often costly or suffer from poor quality. To address these challenges, this paper introduces a framework called EpicPRM, which annotates each intermediate reasoning step based on its quantified contribution and uses an adaptive binary search algorithm to enhance both annotation precision and efficiency. Using this approach, we efficiently construct a high-quality process supervision training dataset named Epic50k, consisting of 50k annotated intermediate steps. Compared to other publicly available datasets, the PRM trained on Epic50k demonstrates significantly superior performance. Getting Epic50k at https://github.com/xiaolizh1/EpicPRM.
☆ MedEthicEval: Evaluating Large Language Models Based on Chinese Medical Ethics
Large language models (LLMs) demonstrate significant potential in advancing medical applications, yet their capabilities in addressing medical ethics challenges remain underexplored. This paper introduces MedEthicEval, a novel benchmark designed to systematically evaluate LLMs in the domain of medical ethics. Our framework encompasses two key components: knowledge, assessing the models' grasp of medical ethics principles, and application, focusing on their ability to apply these principles across diverse scenarios. To support this benchmark, we consulted with medical ethics researchers and developed three datasets addressing distinct ethical challenges: blatant violations of medical ethics, priority dilemmas with clear inclinations, and equilibrium dilemmas without obvious resolutions. MedEthicEval serves as a critical tool for understanding LLMs' ethical reasoning in healthcare, paving the way for their responsible and effective use in medical contexts.
☆ Iterative Value Function Optimization for Guided Decoding
While Reinforcement Learning from Human Feedback (RLHF) has become the predominant method for controlling language model outputs, it suffers from high computational costs and training instability. Guided decoding, especially value-guided methods, offers a cost-effective alternative by controlling outputs without re-training models. However, the accuracy of the value function is crucial for value-guided decoding, as inaccuracies can lead to suboptimal decision-making and degraded performance. Existing methods struggle with accurately estimating the optimal value function, leading to less effective control. We propose Iterative Value Function Optimization, a novel framework that addresses these limitations through two key components: Monte Carlo Value Estimation, which reduces estimation variance by exploring diverse trajectories, and Iterative On-Policy Optimization, which progressively improves value estimation through collecting trajectories from value-guided policies. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of value-guided decoding approaches in aligning language models. These approaches not only achieve alignment but also significantly reduce computational costs by leveraging principled value function optimization for efficient and effective control.
comment: 20 pages, 10 figures
☆ EchoQA: A Large Collection of Instruction Tuning Data for Echocardiogram Reports NeurIPS
We introduce a novel question-answering (QA) dataset using echocardiogram reports sourced from the Medical Information Mart for Intensive Care database. This dataset is specifically designed to enhance QA systems in cardiology, consisting of 771,244 QA pairs addressing a wide array of cardiac abnormalities and their severity. We compare large language models (LLMs), including open-source and biomedical-specific models for zero-shot evaluation, and closed-source models for zero-shot and three-shot evaluation. Our results show that fine-tuning LLMs improves performance across various QA metrics, validating the value of our dataset. Clinicians also qualitatively evaluate the best-performing model to assess the LLM responses for correctness. Further, we conduct fine-grained fairness audits to assess the bias-performance trade-off of LLMs across various social determinants of health. Our objective is to propel the field forward by establishing a benchmark for LLM AI agents aimed at supporting clinicians with cardiac differential diagnoses, thereby reducing the documentation burden that contributes to clinician burnout and enabling healthcare professionals to focus more on patient care.
comment: NeurIPS SafeGenAI 2024
☆ Add-One-In: Incremental Sample Selection for Large Language Models via a Choice-Based Greedy Paradigm
Selecting high-quality and diverse training samples from extensive datasets plays a crucial role in reducing training overhead and enhancing the performance of Large Language Models (LLMs). However, existing studies fall short in assessing the overall value of selected data, focusing primarily on individual quality, and struggle to strike an effective balance between ensuring diversity and minimizing data point traversals. Therefore, this paper introduces a novel choice-based sample selection framework that shifts the focus from evaluating individual sample quality to comparing the contribution value of different samples when incorporated into the subset. Thanks to the advanced language understanding capabilities of LLMs, we utilize LLMs to evaluate the value of each option during the selection process. Furthermore, we design a greedy sampling process where samples are incrementally added to the subset, thereby improving efficiency by eliminating the need for exhaustive traversal of the entire dataset with the limited budget. Extensive experiments demonstrate that selected data from our method not only surpass the performance of the full dataset but also achieves competitive results with state-of-the-art (SOTA) studies, while requiring fewer selections. Moreover, we validate our approach on a larger medical dataset, highlighting its practical applicability in real-world applications.
☆ Are Large Vision Language Models Good Game Players? ICLR2025
Large Vision Language Models (LVLMs) have demonstrated remarkable abilities in understanding and reasoning about both visual and textual information. However, existing evaluation methods for LVLMs, primarily based on benchmarks like Visual Question Answering and image captioning, often fail to capture the full scope of LVLMs' capabilities. These benchmarks are limited by issues such as inadequate assessment of detailed visual perception, data contamination, and a lack of focus on multi-turn reasoning. To address these challenges, we propose \method{}, a game-based evaluation framework designed to provide a comprehensive assessment of LVLMs' cognitive and reasoning skills in structured environments. \method{} uses a set of games to evaluate LVLMs on four core tasks: Perceiving, Question Answering, Rule Following, and End-to-End Playing, with each target task designed to assess specific abilities, including visual perception, reasoning, decision-making, etc. Based on this framework, we conduct extensive experiments that explore the limitations of current LVLMs, such as handling long structured outputs and perceiving detailed and dense elements. Code and data are publicly available at https://github.com/xinke-wang/LVLM-Playground.
comment: ICLR2025
☆ DeLTa: A Decoding Strategy based on Logit Trajectory Prediction Improves Factuality and Reasoning Ability
Large Language Models (LLMs) are increasingly being used in real-world applications. However, concerns about the reliability of the content they generate persist, as it frequently deviates from factual correctness or exhibits deficiencies in logical reasoning. This paper proposes a novel decoding strategy aimed at enhancing both factual accuracy and inferential reasoning without requiring any modifications to the architecture or pre-trained parameters of LLMs. Our approach adjusts next-token probabilities by analyzing the trajectory of logits from lower to higher layers in Transformers and applying linear regression. We find that this Decoding by Logit Trajectory-based approach (DeLTa) effectively reinforces factuality and reasoning while mitigating incorrect generation. Experiments on TruthfulQA demonstrate that DeLTa attains up to a 4.9% improvement over the baseline. Furthermore, it enhances performance by up to 8.1% on StrategyQA and 7.3% on GSM8K, both of which demand strong reasoning capabilities.
comment: Source code is available at https://github.com/githubhyz/DeLTa
☆ Unlocking a New Rust Programming Experience: Fast and Slow Thinking with LLMs to Conquer Undefined Behaviors
To provide flexibility and low-level interaction capabilities, the unsafe tag in Rust is essential in many projects, but undermines memory safety and introduces Undefined Behaviors (UBs) that reduce safety. Eliminating these UBs requires a deep understanding of Rust's safety rules and strong typing. Traditional methods require depth analysis of code, which is laborious and depends on knowledge design. The powerful semantic understanding capabilities of LLM offer new opportunities to solve this problem. Although existing large model debugging frameworks excel in semantic tasks, limited by fixed processes and lack adaptive and dynamic adjustment capabilities. Inspired by the dual process theory of decision-making (Fast and Slow Thinking), we present a LLM-based framework called RustBrain that automatically and flexibly minimizes UBs in Rust projects. Fast thinking extracts features to generate solutions, while slow thinking decomposes, verifies, and generalizes them abstractly. To apply verification and generalization results to solution generation, enabling dynamic adjustments and precise outputs, RustBrain integrates two thinking through a feedback mechanism. Experimental results on Miri dataset show a 94.3% pass rate and 80.4% execution rate, improving flexibility and Rust projects safety.
☆ Examining the Mental Health Impact of Misinformation on Social Media Using a Hybrid Transformer-Based Approach
Social media has significantly reshaped interpersonal communication, fostering connectivity while also enabling the proliferation of misinformation. The unchecked spread of false narratives has profound effects on mental health, contributing to increased stress, anxiety, and misinformation-driven paranoia. This study presents a hybrid transformer-based approach using a RoBERTa-LSTM classifier to detect misinformation, assess its impact on mental health, and classify disorders linked to misinformation exposure. The proposed models demonstrate accuracy rates of 98.4, 87.8, and 77.3 in detecting misinformation, mental health implications, and disorder classification, respectively. Furthermore, Pearson's Chi-Squared Test for Independence (p-value = 0.003871) validates the direct correlation between misinformation and deteriorating mental well-being. This study underscores the urgent need for better misinformation management strategies to mitigate its psychological repercussions. Future research could explore broader datasets incorporating linguistic, demographic, and cultural variables to deepen the understanding of misinformation-induced mental health distress.
comment: 20 pages
☆ Limited Effectiveness of LLM-based Data Augmentation for COVID-19 Misinformation Stance Detection
Misinformation surrounding emerging outbreaks poses a serious societal threat, making robust countermeasures essential. One promising approach is stance detection (SD), which identifies whether social media posts support or oppose misleading claims. In this work, we finetune classifiers on COVID-19 misinformation SD datasets consisting of claims and corresponding tweets. Specifically, we test controllable misinformation generation (CMG) using large language models (LLMs) as a method for data augmentation. While CMG demonstrates the potential for expanding training datasets, our experiments reveal that performance gains over traditional augmentation methods are often minimal and inconsistent, primarily due to built-in safeguards within LLMs. We release our code and datasets to facilitate further research on misinformation detection and generation.
PromptCoT: Synthesizing Olympiad-level Problems for Mathematical Reasoning in Large Language Models
The ability of large language models to solve complex mathematical problems has progressed significantly, particularly for tasks requiring advanced reasoning. However, the scarcity of sufficiently challenging problems, particularly at the Olympiad level, hinders further advancements. In this work, we introduce PromptCoT, a novel approach for automatically generating high-quality Olympiad-level math problems. The proposed method synthesizes complex problems based on mathematical concepts and the rationale behind problem construction, emulating the thought processes of experienced problem designers. We provide a theoretical analysis demonstrating that an optimal rationale should maximize both the likelihood of rationale generation given the associated concepts and the likelihood of problem generation conditioned on both the rationale and the concepts. Our method is evaluated on standard benchmarks including GSM8K, MATH-500, and AIME2024, where it consistently outperforms existing problem generation methods. Furthermore, we demonstrate that PromptCoT exhibits superior data scalability, consistently maintaining high performance as the dataset size increases, outperforming the baselines. The implementation is available at https://github.com/zhaoxlpku/PromptCoT.
comment: Preprint
☆ Audio-Reasoner: Improving Reasoning Capability in Large Audio Language Models
Recent advancements in multimodal reasoning have largely overlooked the audio modality. We introduce Audio-Reasoner, a large-scale audio language model for deep reasoning in audio tasks. We meticulously curated a large-scale and diverse multi-task audio dataset with simple annotations. Then, we leverage closed-source models to conduct secondary labeling, QA generation, along with structured COT process. These datasets together form a high-quality reasoning dataset with 1.2 million reasoning-rich samples, which we name CoTA. Following inference scaling principles, we train Audio-Reasoner on CoTA, enabling it to achieve great logical capabilities in audio reasoning. Experiments show state-of-the-art performance across key benchmarks, including MMAU-mini (+25.42%), AIR-Bench chat/foundation(+14.57%/+10.13%), and MELD (+8.01%). Our findings stress the core of structured CoT training in advancing audio reasoning.
comment: Technical report, in process
☆ AxBERT: An Interpretable Chinese Spelling Correction Method Driven by Associative Knowledge Network
Deep learning has shown promising performance on various machine learning tasks. Nevertheless, the uninterpretability of deep learning models severely restricts the usage domains that require feature explanations, such as text correction. Therefore, a novel interpretable deep learning model (named AxBERT) is proposed for Chinese spelling correction by aligning with an associative knowledge network (AKN). Wherein AKN is constructed based on the co-occurrence relations among Chinese characters, which denotes the interpretable statistic logic contrasted with uninterpretable BERT logic. And a translator matrix between BERT and AKN is introduced for the alignment and regulation of the attention component in BERT. In addition, a weight regulator is designed to adjust the attention distributions in BERT to appropriately model the sentence semantics. Experimental results on SIGHAN datasets demonstrate that AxBERT can achieve extraordinary performance, especially upon model precision compared to baselines. Our interpretable analysis, together with qualitative reasoning, can effectively illustrate the interpretability of AxBERT.
☆ OmniSQL: Synthesizing High-quality Text-to-SQL Data at Scale
Text-to-SQL, the task of translating natural language questions into SQL queries, plays a crucial role in enabling non-experts to interact with databases. While recent advancements in large language models (LLMs) have significantly enhanced text-to-SQL performance, existing approaches face notable limitations in real-world text-to-SQL applications. Prompting-based methods often depend on closed-source LLMs, which are expensive, raise privacy concerns, and lack customization. Fine-tuning-based methods, on the other hand, suffer from poor generalizability due to the limited coverage of publicly available training data. To overcome these challenges, we propose a novel and scalable text-to-SQL data synthesis framework for automatically synthesizing large-scale, high-quality, and diverse datasets without extensive human intervention. Using this framework, we introduce SynSQL-2.5M, the first million-scale text-to-SQL dataset, containing 2.5 million samples spanning over 16,000 synthetic databases. Each sample includes a database, SQL query, natural language question, and chain-of-thought (CoT) solution. Leveraging SynSQL-2.5M, we develop OmniSQL, a powerful open-source text-to-SQL model available in three sizes: 7B, 14B, and 32B. Extensive evaluations across nine datasets demonstrate that OmniSQL achieves state-of-the-art performance, matching or surpassing leading closed-source and open-source LLMs, including GPT-4o and DeepSeek-V3, despite its smaller size. We release all code, datasets, and models to support further research.
☆ Haste Makes Waste: Evaluating Planning Abilities of LLMs for Efficient and Feasible Multitasking with Time Constraints Between Actions
While Large Language Model-based agents have demonstrated substantial progress in task completion, existing evaluation benchmarks tend to overemphasize single-task performance, with insufficient attention given to the crucial aspects of multitask planning and execution efficiency required in real-world scenarios. To bridge this gap, we present Recipe2Plan, a novel benchmark framework based on real-world cooking scenarios. Unlike conventional benchmarks, Recipe2Plan challenges agents to optimize cooking time through parallel task execution while respecting temporal constraints i.e. specific actions need to be performed within a particular time intervals following the preceding steps. Overly aggressive local parallelization may disrupt this constraint, potentially compromising the entire cooking process. This strict time constraint between actions raises a unique challenge for agents to balance between maximizing concurrent operations and adhering to critical timing constraints. Extensive experiments with state-of-the-art models reveal challenges in maintaining this balance between efficiency and feasibility. The results highlight the need for improved temporal awareness and global multitasking capabilities in large language models. We open-source our benchmark and code at https://github.com/WilliamZR/Recipe2Plan.
☆ Enhancing LLM Reliability via Explicit Knowledge Boundary Modeling
Large language models (LLMs) frequently hallucinate due to misaligned self-awareness, generating erroneous outputs when addressing queries beyond their knowledge boundaries. While existing approaches mitigate hallucinations via uncertainty estimation or query rejection, they suffer from computational inefficiency or sacrificed helpfulness. To address these issues, we propose the Explicit Knowledge Boundary Modeling (EKBM) framework, integrating fast and slow reasoning systems to harmonize reliability and usability. The framework first employs a fast-thinking model to generate confidence-labeled responses, enabling immediate use of high-confidence outputs. For uncertain predictions, a slow refinement model conducts targeted reasoning to improve accuracy. To align model behavior with our proposed object, we propose a hybrid training pipeline, enhancing self-awareness without degrading task performance. Evaluations on dialogue state tracking tasks demonstrate that EKBM achieves superior model reliability over uncertainty-based baselines. Further analysis reveals that refinement substantially boosts accuracy while maintaining low computational overhead. Our work establishes a scalable paradigm for advancing LLM reliability and balancing accuracy and practical utility in error-sensitive applications.
☆ Words or Vision: Do Vision-Language Models Have Blind Faith in Text? CVPR 2025
Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks, but their handling of inconsistencies between modalities is underexplored. We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings. By introducing textual variations to four vision-centric tasks and evaluating ten Vision-Language Models (VLMs), we discover a \emph{``blind faith in text''} phenomenon: VLMs disproportionately trust textual data over visual data when inconsistencies arise, leading to significant performance drops under corrupted text and raising safety concerns. We analyze factors influencing this text bias, including instruction prompts, language model size, text relevance, token order, and the interplay between visual and textual certainty. While certain factors, such as scaling up the language model size, slightly mitigate text bias, others like token order can exacerbate it due to positional biases inherited from language models. To address this issue, we explore supervised fine-tuning with text augmentation and demonstrate its effectiveness in reducing text bias. Additionally, we provide a theoretical analysis suggesting that the blind faith in text phenomenon may stem from an imbalance of pure text and multi-modal data during training. Our findings highlight the need for balanced training and careful consideration of modality interactions in VLMs to enhance their robustness and reliability in handling multi-modal data inconsistencies.
comment: Accepted to CVPR 2025
☆ ATLaS: Agent Tuning via Learning Critical Steps
Large Language Model (LLM) agents have demonstrated remarkable generalization capabilities across multi-domain tasks. Existing agent tuning approaches typically employ supervised finetuning on entire expert trajectories. However, behavior-cloning of full trajectories can introduce expert bias and weaken generalization to states not covered by the expert data. Additionally, critical steps, such as planning, complex reasoning for intermediate subtasks, and strategic decision-making, are essential to success in agent tasks, so learning these steps is the key to improving LLM agents. For more effective and efficient agent tuning, we propose ATLaS that identifies the critical steps in expert trajectories and finetunes LLMs solely on these steps with reduced costs. By steering the training's focus to a few critical steps, our method mitigates the risk of overfitting entire trajectories and promotes generalization across different environments and tasks. In extensive experiments, an LLM finetuned on only 30% critical steps selected by ATLaS outperforms the LLM finetuned on all steps and recent open-source LLM agents. ATLaS maintains and improves base LLM skills as generalist agents interacting with diverse environments.
☆ Adversarial Tokenization
Current LLM pipelines account for only one possible tokenization for a given string, ignoring exponentially many alternative tokenizations during training and inference. For example, the standard Llama3 tokenization of penguin is [p,enguin], yet [peng,uin] is another perfectly valid alternative. In this paper, we show that despite LLMs being trained solely on one tokenization, they still retain semantic understanding of other tokenizations, raising questions about their implications in LLM safety. Put succinctly, we answer the following question: can we adversarially tokenize an obviously malicious string to evade safety and alignment restrictions? We show that not only is adversarial tokenization an effective yet previously neglected axis of attack, but it is also competitive against existing state-of-the-art adversarial approaches without changing the text of the harmful request. We empirically validate this exploit across three state-of-the-art LLMs and adversarial datasets, revealing a previously unknown vulnerability in subword models.
☆ Tabby: Tabular Data Synthesis with Language Models
While advances in large language models (LLMs) have greatly improved the quality of synthetic text data in recent years, synthesizing tabular data has received relatively less attention. We address this disparity with Tabby, a simple but powerful post-training modification to the standard Transformer language model architecture, enabling its use for tabular dataset synthesis. Tabby enables the representation of differences across columns using Gated Mixture-of-Experts, with column-specific sets of parameters. Empirically, Tabby results in data quality near or equal to that of real data. By pairing our novel LLM table training technique, Plain, with Tabby, we observe up to a 44% improvement in quality over previous methods. We also show that Tabby extends beyond tables to more general structured data, reaching parity with real data on a nested JSON dataset as well.
comment: 21 pages, 8 figures
☆ Malware Classification from Memory Dumps Using Machine Learning, Transformers, and Large Language Models
This study investigates the performance of various classification models for a malware classification task using different feature sets and data configurations. Six models-Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Decision Trees, Random Forest (RF), and Extreme Gradient Boosting (XGB)-were evaluated alongside two deep learning models, Recurrent Neural Networks (RNN) and Transformers, as well as the Gemini zero-shot and few-shot learning methods. Four feature sets were tested including All Features, Literature Review Features, the Top 45 Features from RF, and Down-Sampled with Top 45 Features. XGB achieved the highest accuracy of 87.42% using the Top 45 Features, outperforming all other models. RF followed closely with 87.23% accuracy on the same feature set. In contrast, deep learning models underperformed, with RNN achieving 66.71% accuracy and Transformers reaching 71.59%. Down-sampling reduced performance across all models, with XGB dropping to 81.31%. Gemini zero-shot and few-shot learning approaches showed the lowest performance, with accuracies of 40.65% and 48.65%, respectively. The results highlight the importance of feature selection in improving model performance while reducing computational complexity. Traditional models like XGB and RF demonstrated superior performance, while deep learning and few-shot methods struggled to match their accuracy. This study underscores the effectiveness of traditional machine learning models for structured datasets and provides a foundation for future research into hybrid approaches and larger datasets.
☆ Measuring Intrinsic Dimension of Token Embeddings
In this study, we measure the Intrinsic Dimension (ID) of token embedding to estimate the intrinsic dimensions of the manifolds spanned by the representations, so as to evaluate their redundancy quantitatively compared to their extrinsic dimensionality. In detail, (1) we estimate the ID of token embeddings in small-scale language models and also modern large language models, finding that the embedding spaces often reside on lower-dimensional manifolds compared to their extrinsic dimensionality; (2) we measure the ID across various model sizes and observe an increase in redundancy rates as the model scale grows; (3) we measure the dynamics of IDs during the training process, and find a rapid ID drop in the early stages of training. Moreover, (4) when LoRA is applied to the embedding layers, we observe a sudden drop in perplexity around the estimated IDs, suggesting that the ID can serve as a useful guideline for LoRA application.
comment: 4 pages, 4 figures
☆ Network Traffic Classification Using Machine Learning, Transformer, and Large Language Models
This study uses various models to address network traffic classification, categorizing traffic into web, browsing, IPSec, backup, and email. We collected a comprehensive dataset from Arbor Edge Defender (AED) devices, comprising of 30,959 observations and 19 features. Multiple models were evaluated, including Naive Bayes, Decision Tree, Random Forest, Gradient Boosting, XGBoost, Deep Neural Networks (DNN), Transformer, and two Large Language Models (LLMs) including GPT-4o and Gemini with zero- and few-shot learning. Transformer and XGBoost showed the best performance, achieving the highest accuracy of 98.95 and 97.56%, respectively. GPT-4o and Gemini showed promising results with few-shot learning, improving accuracy significantly from initial zero-shot performance. While Gemini Few-Shot and GPT-4o Few-Shot performed well in categories like Web and Email, misclassifications occurred in more complex categories like IPSec and Backup. The study highlights the importance of model selection, fine-tuning, and the balance between training data size and model complexity for achieving reliable classification results.
☆ Improving LLM-as-a-Judge Inference with the Judgment Distribution
Using language models to scalably approximate human preferences on text quality (LLM-as-a-judge) has become a standard practice applicable to many tasks. A judgment is often extracted from the judge's textual output alone, typically with greedy decoding. However, LLM judges naturally provide distributions over judgment tokens, inviting a breadth of inference methods for extracting fine-grained preferences. We find that taking the mean of the judgment distribution consistently outperforms taking the mode (i.e. greedy decoding) in all evaluation settings (i.e. pointwise, pairwise, and listwise). We further explore novel methods of deriving preferences from judgment distributions, and find that methods incorporating risk aversion often improve performance. Lastly, we analyze LLM-as-a-judge paired with chain-of-thought (CoT) prompting, showing that CoT can collapse the spread of the judgment distribution, often harming performance. Our findings suggest leveraging distributional output can improve LLM-as-a-judge, as opposed to using the text interface alone.
☆ Semi-Supervised In-Context Learning: A Baseline Study
Most existing work in data selection for In-Context Learning (ICL) has focused on constructing demonstrations from ground truth annotations, with limited attention given to selecting reliable self-generated annotations. In this work, we propose a three-step semi-supervised ICL framework: annotation generation, demonstration selection, and semi-supervised inference. Our baseline, Naive-SemiICL, which prompts select high-confidence self-generated demonstrations for ICL prompting, outperforms a 16-shot baseline by an average of 9.94% across 16 datasets. We further introduce IterPSD, an annotation approach that refines pseudo-demonstrations iteratively, achieving up to 6.8% additional gains in classification tasks. Lastly, we reveal a scaling law for semi-supervised ICL, where models achieve optimal performance with over 1,000 demonstrations.
☆ QE4PE: Word-level Quality Estimation for Human Post-Editing
Word-level quality estimation (QE) detects erroneous spans in machine translations, which can direct and facilitate human post-editing. While the accuracy of word-level QE systems has been assessed extensively, their usability and downstream influence on the speed, quality and editing choices of human post-editing remain understudied. Our QE4PE study investigates the impact of word-level QE on machine translation (MT) post-editing in a realistic setting involving 42 professional post-editors across two translation directions. We compare four error-span highlight modalities, including supervised and uncertainty-based word-level QE methods, for identifying potential errors in the outputs of a state-of-the-art neural MT model. Post-editing effort and productivity are estimated by behavioral logs, while quality improvements are assessed by word- and segment-level human annotation. We find that domain, language and editors' speed are critical factors in determining highlights' effectiveness, with modest differences between human-made and automated QE highlights underlining a gap between accuracy and usability in professional workflows.
comment: Code: https://github.com/gsarti/qe4pe. Dataset: https://huggingface.co/datasets/gsarti/qe4pe
☆ SAGE: Steering and Refining Dialog Generation with State-Action Augmentation
Recent advances in large language models have demonstrated impressive capabilities in task-oriented applications, yet building emotionally intelligent chatbots that can engage in natural, strategic conversations remains a challenge. We present a novel approach called SAGE that uses latent variables to control long-horizon behavior in dialogue generation. At the core of our method is the State-Action Chain (SAC), which augments standard language model fine-tuning by introducing latent variables that encapsulate emotional states and conversational strategies between dialogue turns. During inference, these variables are generated before each response, enabling coarse-grained control over dialogue progression while maintaining natural interaction patterns. We also introduce a self-improvement pipeline that leverages dialogue tree search, LLM-based reward modeling, and targeted fine-tuning to optimize conversational trajectories. Our experimental results show that models trained with this approach demonstrate improved performance in emotional intelligence metrics while maintaining strong capabilities on LLM benchmarks. The discrete nature of our latent variables facilitates search-based strategies and provides a foundation for future applications of reinforcement learning to dialogue systems, where learning can occur at the state level rather than the token level.
☆ SAFE: A Sparse Autoencoder-Based Framework for Robust Query Enrichment and Hallucination Mitigation in LLMs
Despite the state-of-the-art performance of Large Language Models (LLMs), these models often suffer from hallucinations, which can undermine their performance in critical applications. In this work, we propose SAFE, a novel method for detecting and mitigating hallucinations by leveraging Sparse Autoencoders (SAEs). While hallucination detection techniques and SAEs have been explored independently, their synergistic application in a comprehensive system, particularly for hallucination-aware query enrichment, has not been fully investigated. To validate the effectiveness of SAFE, we evaluate it on two models with available SAEs across three diverse cross-domain datasets designed to assess hallucination problems. Empirical results demonstrate that SAFE consistently improves query generation accuracy and mitigates hallucinations across all datasets, achieving accuracy improvements of up to 29.45%.
☆ One Model to Train them All: Hierarchical Self-Distillation for Enhanced Early Layer Embeddings
Deploying language models often requires handling model size vs. performance trade-offs to satisfy downstream latency constraints while preserving the model's usefulness. Model distillation is commonly employed to reduce model size while maintaining acceptable performance. However, distillation can be inefficient since it involves multiple training steps. In this work, we introduce MODULARSTARENCODER, a modular multi-exit encoder with 1B parameters, useful for multiple tasks within the scope of code retrieval. MODULARSTARENCODER is trained with a novel self-distillation mechanism that significantly improves lower-layer representations-allowing different portions of the model to be used while still maintaining a good trade-off in terms of performance. Our architecture focuses on enhancing text-to-code and code-to-code search by systematically capturing syntactic and semantic structures across multiple levels of representation. Specific encoder layers are targeted as exit heads, allowing higher layers to guide earlier layers during training. This self-distillation effect improves intermediate representations, increasing retrieval recall at no extra training cost. In addition to the multi-exit scheme, our approach integrates a repository-level contextual loss that maximally utilizes the training context window, further enhancing the learned representations. We also release a new dataset constructed via code translation, seamlessly expanding traditional text-to-code benchmarks with code-to-code pairs across diverse programming languages. Experimental results highlight the benefits of self-distillation through multi-exit supervision.
☆ Will I Get Hate Speech Predicting the Volume of Abusive Replies before Posting in Social Media
Despite the growing body of research tackling offensive language in social media, this research is predominantly reactive, determining if content already posted in social media is abusive. There is a gap in predictive approaches, which we address in our study by enabling to predict the volume of abusive replies a tweet will receive after being posted. We formulate the problem from the perspective of a social media user asking: ``if I post a certain message on social media, is it possible to predict the volume of abusive replies it might receive?'' We look at four types of features, namely text, text metadata, tweet metadata, and account features, which also help us understand the extent to which the user or the content helps predict the number of abusive replies. This, in turn, helps us develop a model to support social media users in finding the best way to post content. One of our objectives is also to determine the extent to which the volume of abusive replies that a tweet will get are motivated by the content of the tweet or by the identity of the user posting it. Our study finds that one can build a model that performs competitively by developing a comprehensive set of features derived from the content of the message that is going to be posted. In addition, our study suggests that features derived from the user's identity do not impact model performance, hence suggesting that it is especially the content of a post that triggers abusive replies rather than who the user is.
☆ Zero-Shot Multi-Label Classification of Bangla Documents: Large Decoders Vs. Classic Encoders
Bangla, a language spoken by over 300 million native speakers and ranked as the sixth most spoken language worldwide, presents unique challenges in natural language processing (NLP) due to its complex morphological characteristics and limited resources. While recent Large Decoder Based models (LLMs), such as GPT, LLaMA, and DeepSeek, have demonstrated excellent performance across many NLP tasks, their effectiveness in Bangla remains largely unexplored. In this paper, we establish the first benchmark comparing decoder-based LLMs with classic encoder-based models for Zero-Shot Multi-Label Classification (Zero-Shot-MLC) task in Bangla. Our evaluation of 32 state-of-the-art models reveals that, existing so-called powerful encoders and decoders still struggle to achieve high accuracy on the Bangla Zero-Shot-MLC task, suggesting a need for more research and resources for Bangla NLP.
☆ Effectively Steer LLM To Follow Preference via Building Confident Directions
Having an LLM that aligns with human preferences is essential for accommodating individual needs, such as maintaining writing style or generating specific topics of interest. The majority of current alignment methods rely on fine-tuning or prompting, which can be either costly or difficult to control. Model steering algorithms, which modify the model output by constructing specific steering directions, are typically easy to implement and optimization-free. However, their capabilities are typically limited to steering the model into one of the two directions (i.e., bidirectional steering), and there has been no theoretical understanding to guarantee their performance. In this work, we propose a theoretical framework to understand and quantify the model steering methods. Inspired by the framework, we propose a confident direction steering method (CONFST) that steers LLMs via modifying their activations at inference time. More specifically, CONFST builds a confident direction that is closely aligned with users' preferences, and this direction is then added to the activations of the LLMs to effectively steer the model output. Our approach offers three key advantages over popular bidirectional model steering methods: 1) It is more powerful, since multiple (i.e. more than two) users' preferences can be aligned simultaneously; 2) It is simple to implement, since there is no need to determine which layer to add the steering vector to; 3) No explicit user instruction is required. We validate our method on GPT-2 XL (1.5B), Mistral (7B) and Gemma-it (9B) models for tasks that require shifting the output of LLMs across various topics and styles, achieving superior performance over competing methods.
☆ LINGOLY-TOO: Disentangling Memorisation from Reasoning with Linguistic Templatisation and Orthographic Obfuscation
Effective evaluation of the reasoning capabilities of large language models (LLMs) are susceptible to overestimation due to data exposure of evaluation benchmarks. We introduce a framework for producing linguistic reasoning problems that reduces the effect of memorisation in model performance estimates and apply this framework to develop LINGOLY-TOO, a challenging evaluation benchmark for linguistic reasoning. By developing orthographic templates, we dynamically obfuscate the writing systems of real languages to generate numerous question variations. These variations preserve the reasoning steps required for each solution while reducing the likelihood of specific problem instances appearing in model training data. Our experiments demonstrate that frontier models, including OpenAI o1-preview and DeepSeem R1, struggle with advanced reasoning. Our analysis also shows that LLMs exhibit noticeable variance in accuracy across permutations of the same problem, and on average perform better on questions appearing in their original orthography. Our findings highlight the opaque nature of response generation in LLMs and provide evidence that prior data exposure contributes to overestimating the reasoning capabilities of frontier models.
☆ Multilingual Relative Clause Attachment Ambiguity Resolution in Large Language Models ACL
This study examines how large language models (LLMs) resolve relative clause (RC) attachment ambiguities and compares their performance to human sentence processing. Focusing on two linguistic factors, namely the length of RCs and the syntactic position of complex determiner phrases (DPs), we assess whether LLMs can achieve human-like interpretations amid the complexities of language. In this study, we evaluated several LLMs, including Claude, Gemini and Llama, in multiple languages: English, Spanish, French, German, Japanese, and Korean. While these models performed well in Indo-European languages (English, Spanish, French, and German), they encountered difficulties in Asian languages (Japanese and Korean), often defaulting to incorrect English translations. The findings underscore the variability in LLMs' handling of linguistic ambiguities and highlight the need for model improvements, particularly for non-European languages. This research informs future enhancements in LLM design to improve accuracy and human-like processing in diverse linguistic environments.
comment: Accepted at PACLIC 2024
☆ InfiniSST: Simultaneous Translation of Unbounded Speech with Large Language Model
Simultaneous translation of unbounded streaming speech remains a challenging problem due to the need for effectively processing the history speech context and past translations so that quality and latency, including computation overhead, can be balanced. Most prior works assume pre-segmented speech, limiting their real-world applicability. In this paper, we propose InfiniSST, a novel approach that formulates SST as a multi-turn dialogue task, enabling seamless translation of unbounded speech. We construct translation trajectories and robust segments from MuST-C with multi-latency augmentation during training and develop a key-value (KV) cache management strategy to facilitate efficient inference. Experiments on MuST-C En-Es, En-De, and En-Zh demonstrate that InfiniSST reduces computation-aware latency by 0.5 to 1 second while maintaining the same translation quality compared to baselines. Ablation studies further validate the contributions of our data construction and cache management strategy. We release the code at https://github.com/LeiLiLab/InfiniSST
comment: Under Review
☆ KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding
We introduce KodCode, a synthetic dataset that addresses the persistent challenge of acquiring high-quality, verifiable training data across diverse difficulties and domains for training Large Language Models for coding. Existing code-focused resources typically fail to ensure either the breadth of coverage (e.g., spanning simple coding tasks to advanced algorithmic problems) or verifiable correctness (e.g., unit tests). In contrast, KodCode comprises question-solution-test triplets that are systematically validated via a self-verification procedure. Our pipeline begins by synthesizing a broad range of coding questions, then generates solutions and test cases with additional attempts allocated to challenging problems. Finally, post-training data synthesis is done by rewriting questions into diverse formats and generating responses under a test-based reject sampling procedure from a reasoning model (DeepSeek R1). This pipeline yields a large-scale, robust and diverse coding dataset. KodCode is suitable for supervised fine-tuning and the paired unit tests also provide great potential for RL tuning. Fine-tuning experiments on coding benchmarks (HumanEval(+), MBPP(+), BigCodeBench, and LiveCodeBench) demonstrate that KodCode-tuned models achieve state-of-the-art performance, surpassing models like Qwen2.5-Coder-32B-Instruct and DeepSeek-R1-Distill-Llama-70B.
comment: Codes and Data: https://kodcode-ai.github.io/
☆ LiteWebAgent: The Open-Source Suite for VLM-Based Web-Agent Applications
We introduce LiteWebAgent, an open-source suite for VLM-based web agent applications. Our framework addresses a critical gap in the web agent ecosystem with a production-ready solution that combines minimal serverless backend configuration, intuitive user and browser interfaces, and extensible research capabilities in agent planning, memory, and tree search. For the core LiteWebAgent agent framework, we implemented a simple yet effective baseline using recursive function calling, providing with decoupled action generation and action grounding. In addition, we integrate advanced research components such as agent planning, agent workflow memory, and tree search in a modular and extensible manner. We then integrate the LiteWebAgent agent framework with frontend and backend as deployed systems in two formats: (1) a production Vercel-based web application, which provides users with an agent-controlled remote browser, (2) a Chrome extension leveraging LiteWebAgent's API to control an existing Chrome browser via CDP (Chrome DevTools Protocol). The LiteWebAgent framework is available at https://github.com/PathOnAI/LiteWebAgent, with deployed frontend at https://lite-web-agent.vercel.app/.
☆ ExpertGenQA: Open-ended QA generation in Specialized Domains
Generating high-quality question-answer pairs for specialized technical domains remains challenging, with existing approaches facing a tradeoff between leveraging expert examples and achieving topical diversity. We present ExpertGenQA, a protocol that combines few-shot learning with structured topic and style categorization to generate comprehensive domain-specific QA pairs. Using U.S. Federal Railroad Administration documents as a test bed, we demonstrate that ExpertGenQA achieves twice the efficiency of baseline few-shot approaches while maintaining $94.4\%$ topic coverage. Through systematic evaluation, we show that current LLM-based judges and reward models exhibit strong bias toward superficial writing styles rather than content quality. Our analysis using Bloom's Taxonomy reveals that ExpertGenQA better preserves the cognitive complexity distribution of expert-written questions compared to template-based approaches. When used to train retrieval models, our generated queries improve top-1 accuracy by $13.02\%$ over baseline performance, demonstrating their effectiveness for downstream applications in technical domains.
♻ ☆ Speculative Decoding and Beyond: An In-Depth Survey of Techniques
Sequential dependencies present a fundamental bottleneck in deploying large-scale autoregressive models, particularly for real-time applications. While traditional optimization approaches like pruning and quantization often compromise model quality, recent advances in generation-refinement frameworks demonstrate that this trade-off can be significantly mitigated. This survey presents a comprehensive taxonomy of generation-refinement frameworks, analyzing methods across autoregressive sequence tasks. We categorize methods based on their generation strategies (from simple n-gram prediction to sophisticated draft models) and refinement mechanisms (including single-pass verification and iterative approaches). Through systematic analysis of both algorithmic innovations and system-level implementations, we examine deployment strategies across computing environments and explore applications spanning text, images, and speech generation. This systematic examination of both theoretical frameworks and practical implementations provides a foundation for future research in efficient autoregressive decoding.
♻ ☆ What is Wrong with Perplexity for Long-context Language Modeling?
Handling long-context inputs is crucial for large language models (LLMs) in tasks such as extended conversations, document summarization, and many-shot in-context learning. While recent approaches have extended the context windows of LLMs and employed perplexity (PPL) as a standard evaluation metric, PPL has proven unreliable for assessing long-context capabilities. The underlying cause of this limitation has remained unclear. In this work, we provide a comprehensive explanation for this issue. We find that PPL overlooks key tokens, which are essential for long-context understanding, by averaging across all tokens and thereby obscuring the true performance of models in long-context scenarios. To address this, we propose \textbf{LongPPL}, a novel metric that focuses on key tokens by employing a long-short context contrastive method to identify them. Our experiments demonstrate that LongPPL strongly correlates with performance on various long-context benchmarks (e.g., Pearson correlation of -0.96), significantly outperforming traditional PPL in predictive accuracy. Additionally, we introduce \textbf{LongCE} (Long-context Cross-Entropy) loss, a re-weighting strategy for fine-tuning that prioritizes key tokens, leading to consistent improvements across diverse benchmarks. In summary, these contributions offer deeper insights into the limitations of PPL and present effective solutions for accurately evaluating and enhancing the long-context capabilities of LLMs. Code is available at https://github.com/PKU-ML/LongPPL.
♻ ☆ Variational Best-of-N Alignment ICLR 2025
Best-of-N (BoN) is a popular and effective algorithm for aligning language models to human preferences. The algorithm works as follows: at inference time, N samples are drawn from the language model, and the sample with the highest reward, as judged by a reward model, is returned as the output. Despite its effectiveness, BoN is computationally expensive; it reduces sampling throughput by a factor of N. To make BoN more efficient at inference time, one strategy is to fine-tune the language model to mimic what BoN does during inference. To achieve this, we derive the distribution induced by the BoN algorithm. We then propose to fine-tune the language model to minimize backward KL divergence to the BoN distribution. Our approach is analogous to mean-field variational inference and, thus, we term it variational BoN (vBoN). To the extent this fine-tuning is successful and we end up with a good approximation, we have reduced the inference cost by a factor of N. Our experiments on controlled generation and summarization tasks show that BoN is the most effective alignment method, and our variational approximation to BoN achieves the closest performance to BoN and surpasses models fine-tuned using the standard KL-constrained RL objective. In the controlled generation task, vBoN appears more frequently on the Pareto frontier of reward and KL divergence compared to other alignment methods. In the summarization task, vBoN achieves high reward values across various sampling temperatures.
comment: Accepted at ICLR 2025
♻ ☆ A-MEM: Agentic Memory for LLM Agents
While large language model (LLM) agents can effectively use external tools for complex real-world tasks, they require memory systems to leverage historical experiences. Current memory systems enable basic storage and retrieval but lack sophisticated memory organization, despite recent attempts to incorporate graph databases. Moreover, these systems' fixed operations and structures limit their adaptability across diverse tasks. To address this limitation, this paper proposes a novel agentic memory system for LLM agents that can dynamically organize memories in an agentic way. Following the basic principles of the Zettelkasten method, we designed our memory system to create interconnected knowledge networks through dynamic indexing and linking. When a new memory is added, we generate a comprehensive note containing multiple structured attributes, including contextual descriptions, keywords, and tags. The system then analyzes historical memories to identify relevant connections, establishing links where meaningful similarities exist. Additionally, this process enables memory evolution - as new memories are integrated, they can trigger updates to the contextual representations and attributes of existing historical memories, allowing the memory network to continuously refine its understanding. Our approach combines the structured organization principles of Zettelkasten with the flexibility of agent-driven decision making, allowing for more adaptive and context-aware memory management. Empirical experiments on six foundation models show superior improvement against existing SOTA baselines. The source code for evaluating performance is available at https://github.com/WujiangXu/AgenticMemory, while the source code of agentic memory system is available at https://github.com/agiresearch/A-mem.
♻ ☆ A Survey on Vision-Language-Action Models for Embodied AI
Embodied AI is widely recognized as a key element of artificial general intelligence because it involves controlling embodied agents to perform tasks in the physical world. Building on the success of large language models and vision-language models, a new category of multimodal models -- referred to as vision-language-action models (VLAs) -- has emerged to address language-conditioned robotic tasks in embodied AI by leveraging their distinct ability to generate actions. In recent years, a myriad of VLAs have been developed, making it imperative to capture the rapidly evolving landscape through a comprehensive survey. To this end, we present the first survey on VLAs for embodied AI. This work provides a detailed taxonomy of VLAs, organized into three major lines of research. The first line focuses on individual components of VLAs. The second line is dedicated to developing control policies adept at predicting low-level actions. The third line comprises high-level task planners capable of decomposing long-horizon tasks into a sequence of subtasks, thereby guiding VLAs to follow more general user instructions. Furthermore, we provide an extensive summary of relevant resources, including datasets, simulators, and benchmarks. Finally, we discuss the challenges faced by VLAs and outline promising future directions in embodied AI. We have created a project associated with this survey, which is available at https://github.com/yueen-ma/Awesome-VLA.
comment: Project page: https://github.com/yueen-ma/Awesome-VLA
♻ ☆ Can AI writing be salvaged? Mitigating Idiosyncrasies and Improving Human-AI Alignment in the Writing Process through Edits
LLM-based applications are helping people write, and LLM-generated text is making its way into social media, journalism, and our classrooms. However, the differences between LLM-generated and human written text remain unclear. To explore this, we hired professional writers to edit paragraphs in several creative domains. We first found these writers agree on undesirable idiosyncrasies in LLM generated text, formalizing it into a seven-category taxonomy (e.g. clich\'es, unnecessary exposition). Second, we curated the LAMP corpus: 1,057 LLM-generated paragraphs edited by professional writers according to our taxonomy. Analysis of LAMP reveals that none of the LLMs used in our study (GPT4o, Claude-3.5-Sonnet, Llama-3.1-70b) outperform each other in terms of writing quality, revealing common limitations across model families. Third, building on existing work in automatic editing we evaluated methods to improve LLM-generated text. A large-scale preference annotation confirms that although experts largely prefer text edited by other experts, automatic editing methods show promise in improving alignment between LLM-generated and human-written text.
comment: ACM CHI 2025
♻ ☆ VOILA: Evaluation of MLLMs For Perceptual Understanding and Analogical Reasoning ICLR 2025
Multimodal Large Language Models (MLLMs) have become a powerful tool for integrating visual and textual information. Despite their exceptional performance on visual understanding benchmarks, measuring their ability to reason abstractly across multiple images remains a significant challenge. To address this, we introduce VOILA, a large-scale, open-ended, dynamic benchmark designed to evaluate MLLMs' perceptual understanding and abstract relational reasoning. VOILA employs an analogical mapping approach in the visual domain, requiring models to generate an image that completes an analogy between two given image pairs, reference and application, without relying on predefined choices. Our experiments demonstrate that the analogical reasoning tasks in VOILA present a challenge to MLLMs. Through multi-step analysis, we reveal that current MLLMs struggle to comprehend inter-image relationships and exhibit limited capabilities in high-level relational reasoning. Notably, we observe that performance improves when following a multi-step strategy of least-to-most prompting. Comprehensive evaluations on open-source models and GPT-4o show that on text-based answers, the best accuracy for challenging scenarios is 13% (LLaMa 3.2) and even for simpler tasks is only 29% (GPT-4o), while human performance is significantly higher at 70% across both difficulty levels.
comment: Accepted at ICLR 2025. Code and data: https://github.com/nlylmz/Voila
♻ ☆ LABIIUM: AI-Enhanced Zero-configuration Measurement Automation System
The complexity of laboratory environments requires solutions that simplify instrument interaction and enhance measurement automation. Traditional tools often require configuration, software, and programming skills, creating barriers to productivity. Previous approaches, including dedicated software suites and custom scripts, frequently fall short in providing user-friendly solutions that align with programming practices. We present LABIIUM, an AI-enhanced, zero-configuration measurement automation system designed to streamline experimental workflows and improve user productivity. LABIIUM integrates an AI assistant powered by Large Language Models (LLMs) to generate code. LABIIUM's Lab-Automation-Measurement Bridges (LAMBs) enable seamless instrument connectivity using standard tools such as VSCode and Python, eliminating setup overhead. To demonstrate its capabilities, we conducted experiments involving the measurement of the parametric transfer curve of a simple two-transistor inverting amplifier with a current source load. The AI assistant was evaluated using different prompt scenarios and compared with multiple models, including Claude Sonnet 3.5, Gemini Pro 1.5, and GPT-4o. An expert solution implementing the Gradient-Weighted Adaptive Stochastic Sampling (GWASS) method was used as a baseline. The solutions generated by the AI assistant were compared with the expert solution and a uniform linear sweep baseline with 10,000 points. The graph results show that the LLMs were able to successfully complete the most basic uniform sweep, but LLMs were unable to develop adaptive sweeping algorithms to compete with GWASS. The evaluation underscores LABIIUM's ability to enhance laboratory productivity and support digital transformation in research and industry, and emphasizes the future work required to improve LLM performance in Electronic Measurement Science Tasks.
comment: accepted for IEEE I2MTC 2025
♻ ☆ Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Models ICLR 2025
Model merging, such as model souping, is the practice of combining different models with the same architecture together without further training. In this work, we present a model merging methodology that addresses the difficulty of fine-tuning Large Language Models (LLMs) for target tasks in non-English languages, where task-specific data is often unavailable. We focus on mathematical reasoning and without in-language math data, facilitate cross-lingual transfer by composing language and math capabilities. Starting from the same pretrained model, we fine-tune separate "experts" on math instruction data in English and on generic instruction data in the target language. We then replace the top and bottom transformer layers of the math expert directly with layers from the language expert, which consequently enhances math performance in the target language. The resulting merged models outperform the individual experts and other merging methods on the math benchmark, MGSM, by 10% across four major languages where math instruction data is scarce. In addition, this layer swapping is simple, inexpensive, and intuitive, as it is based on an interpretative analysis of the most important parameter changes during the fine-tuning of each expert. The ability to successfully re-compose LLMs for cross-lingual transfer in this manner opens up future possibilities to combine model expertise, create modular solutions, and transfer reasoning capabilities across languages all post hoc.
comment: ICLR 2025, Spotlight Paper, In The Thirteenth International Conference on Learning Representations, 2025
♻ ☆ Verbalized Probabilistic Graphical Modeling
Human cognition excels at transcending sensory input and forming latent representations that structure our understanding of the world. Although Large Language Models (LLMs) can produce chain-of-thought reasoning, they lack a principled framework to capture latent structures and model uncertainty, especially in compositional reasoning tasks. We propose Verbalized Probabilistic Graphical Modeling (vPGM), a Bayesian prompting framework that guides LLMs to simulate key principles of Probabilistic Graphical Models (PGMs) in natural language. Unlike many traditional probabilistic methods requiring substantial domain expertise or specialized training, vPGM bypasses expert-driven model design, making it well-suited for scenarios with limited assumptions or scarce data. We evaluated our model on several compositional reasoning tasks, both close-ended and open-ended. Our results indicate that the model effectively enhances confidence calibration and text generation quality.
♻ ☆ Why Is Spatial Reasoning Hard for VLMs? An Attention Mechanism Perspective on Focus Areas
Large Vision Language Models (VLMs) have long struggled with spatial reasoning tasks. Surprisingly, even simple spatial reasoning tasks, such as recognizing "under" or "behind" relationships between only two objects, pose significant challenges for current VLMs. In this work, we study the spatial reasoning challenge from the lens of mechanistic interpretability, diving into the model's internal states to examine the interactions between image and text tokens. By tracing attention distribution over the image through out intermediate layers, we observe that successful spatial reasoning correlates strongly with the model's ability to align its attention distribution with actual object locations, particularly differing between familiar and unfamiliar spatial relationships. Motivated by these findings, we propose ADAPTVIS based on inference-time confidence scores to sharpen the attention on highly relevant regions when confident, while smoothing and broadening the attention window to consider a wider context when confidence is lower. This training-free decoding method shows significant improvement (e.g., up to a 50 absolute point improvement) on spatial reasoning benchmarks such as WhatsUp and VSR with negligible cost. We make code and data publicly available for research purposes at https://github.com/shiqichen17/AdaptVis.
♻ ☆ Semantic Volume: Quantifying and Detecting both External and Internal Uncertainty in LLMs
Large language models (LLMs) have demonstrated remarkable performance across diverse tasks by encoding vast amounts of factual knowledge. However, they are still prone to hallucinations, generating incorrect or misleading information, often accompanied by high uncertainty. Existing methods for hallucination detection primarily focus on quantifying internal uncertainty, which arises from missing or conflicting knowledge within the model. However, hallucinations can also stem from external uncertainty, where ambiguous user queries lead to multiple possible interpretations. In this work, we introduce Semantic Volume, a novel mathematical measure for quantifying both external and internal uncertainty in LLMs. Our approach perturbs queries and responses, embeds them in a semantic space, and computes the determinant of the Gram matrix of the embedding vectors, capturing their dispersion as a measure of uncertainty. Our framework provides a generalizable and unsupervised uncertainty detection method without requiring white-box access to LLMs. We conduct extensive experiments on both external and internal uncertainty detection, demonstrating that our Semantic Volume method consistently outperforms existing baselines in both tasks. Additionally, we provide theoretical insights linking our measure to differential entropy, unifying and extending previous sampling-based uncertainty measures such as the semantic entropy. Semantic Volume is shown to be a robust and interpretable approach to improving the reliability of LLMs by systematically detecting uncertainty in both user queries and model responses.
comment: This paper needs approval from Amazon for open resource release
♻ ☆ Revealing the Pragmatic Dilemma for Moral Reasoning Acquisition in Language Models
Ensuring that Large Language Models (LLMs) return just responses which adhere to societal values is crucial for their broader application. Prior research has shown that LLMs often fail to perform satisfactorily on tasks requiring moral cognizance, such as ethics-based judgments. While current approaches have focused on fine-tuning LLMs with curated datasets to improve their capabilities on such tasks, choosing the optimal learning paradigm to enhance the ethical responses of LLMs remains an open research debate. In this work, we aim to address this fundamental question: can current learning paradigms enable LLMs to acquire sufficient moral reasoning capabilities? Drawing from distributional semantics theory and the pragmatic nature of moral discourse, our analysis indicates that performance improvements follow a mechanism similar to that of semantic-level tasks, and therefore remain affected by the pragmatic nature of morals latent in discourse, a phenomenon we name the pragmatic dilemma. We conclude that this pragmatic dilemma imposes significant limitations on the generalization ability of current learning paradigms, making it the primary bottleneck for moral reasoning acquisition in LLMs.
♻ ☆ MAPS: Motivation-Aware Personalized Search via LLM-Driven Consultation Alignment
Personalized product search aims to retrieve and rank items that match users' preferences and search intent. Despite their effectiveness, existing approaches typically assume that users' query fully captures their real motivation. However, our analysis of a real-world e-commerce platform reveals that users often engage in relevant consultations before searching, indicating they refine intents through consultations based on motivation and need. The implied motivation in consultations is a key enhancing factor for personalized search. This unexplored area comes with new challenges including aligning contextual motivations with concise queries, bridging the category-text gap, and filtering noise within sequence history. To address these, we propose a Motivation-Aware Personalized Search (MAPS) method. It embeds queries and consultations into a unified semantic space via LLMs, utilizes a Mixture of Attention Experts (MoAE) to prioritize critical semantics, and introduces dual alignment: (1) contrastive learning aligns consultations, reviews, and product features; (2) bidirectional attention integrates motivation-aware embeddings with user preferences. Extensive experiments on real and synthetic data show MAPS outperforms existing methods in both retrieval and ranking tasks.
♻ ☆ Large Language Models are Powerful EHR Encoders
Electronic Health Records (EHRs) offer rich potential for clinical prediction, yet their inherent complexity and heterogeneity pose significant challenges for traditional machine learning approaches. Domain-specific EHR foundation models trained on large collections of unlabeled EHR data have demonstrated promising improvements in predictive accuracy and generalization; however, their training is constrained by limited access to diverse, high-quality datasets and inconsistencies in coding standards and healthcare practices. In this study, we explore the possibility of using general-purpose Large Language Models (LLMs) based embedding methods as EHR encoders. By serializing patient records into structured Markdown text, transforming codes into human-readable descriptors, we leverage the extensive generalization capabilities of LLMs pretrained on vast public corpora, thereby bypassing the need for proprietary medical datasets. We systematically evaluate two state-of-the-art LLM-embedding models, GTE-Qwen2-7B-Instruct and LLM2Vec-Llama3.1-8B-Instruct, across 15 diverse clinical prediction tasks from the EHRSHOT benchmark, comparing their performance to an EHRspecific foundation model, CLIMBR-T-Base, and traditional machine learning baselines. Our results demonstrate that LLM-based embeddings frequently match or exceed the performance of specialized models, even in few-shot settings, and that their effectiveness scales with the size of the underlying LLM and the available context window. Overall, our findings demonstrate that repurposing LLMs for EHR encoding offers a scalable and effective approach for clinical prediction, capable of overcoming the limitations of traditional EHR modeling and facilitating more interoperable and generalizable healthcare applications.
♻ ☆ Towards Zero-Shot, Controllable Dialog Planning with LLMs AAAI 2022
Recently, Large Language Models (LLMs) have emerged as an alternative to training task-specific dialog agents, due to their broad reasoning capabilities and performance in zero-shot learning scenarios. However, many LLM-based dialog systems fall short in planning towards an overarching dialog goal and therefore cannot steer the conversation appropriately. Furthermore, these models struggle with hallucination, making them unsuitable for information access in sensitive domains, such as legal or medical domains, where correctness of information given to users is critical. The recently introduced task Conversational Tree Search (CTS) proposes the use of dialog graphs to avoid hallucination in sensitive domains, however, state-of-the-art agents are Reinforcement Learning (RL) based and require long training times, despite excelling at dialog strategy. This paper introduces a novel zero-shot method for controllable CTS agents, where LLMs guide the dialog planning through domain graphs by searching and pruning relevant graph nodes based on user interaction preferences. We show that these agents significantly outperform state-of-the-art CTS agents ($p<0.0001$; Barnard Exact test) in simulation. This generalizes to all available CTS domains. Finally, we perform user evaluation to test the agent's performance in the wild, showing that our policy significantly ($p<0.05$; Barnard Exact) improves task-success compared to the state-of-the-art RL-based CTS agent.
comment: This paper has been accepted for publication at the AAAI 2022 Workshop on Planning in the Era of LLMs
♻ ☆ Towards Safe AI Clinicians: A Comprehensive Study on Large Language Model Jailbreaking in Healthcare
Large language models (LLMs) are increasingly utilized in healthcare applications. However, their deployment in clinical practice raises significant safety concerns, including the potential spread of harmful information. This study systematically assesses the vulnerabilities of seven LLMs to three advanced black-box jailbreaking techniques within medical contexts. To quantify the effectiveness of these techniques, we propose an automated and domain-adapted agentic evaluation pipeline. Experiment results indicate that leading commercial and open-source LLMs are highly vulnerable to medical jailbreaking attacks. To bolster model safety and reliability, we further investigate the effectiveness of Continual Fine-Tuning (CFT) in defending against medical adversarial attacks. Our findings underscore the necessity for evolving attack methods evaluation, domain-specific safety alignment, and LLM safety-utility balancing. This research offers actionable insights for advancing the safety and reliability of AI clinicians, contributing to ethical and effective AI deployment in healthcare.
♻ ☆ Annotating and Inferring Compositional Structures in Numeral Systems Across Languages
Numeral systems across the world's languages vary in fascinating ways, both regarding their synchronic structure and the diachronic processes that determined how they evolved in their current shape. For a proper comparison of numeral systems across different languages, however, it is important to code them in a standardized form that allows for the comparison of basic properties. Here, we present a simple but effective coding scheme for numeral annotation, along with a workflow that helps to code numeral systems in a computer-assisted manner, providing sample data for numerals from 1 to 40 in 25 typologically diverse languages. We perform a thorough analysis of the sample, focusing on the systematic comparison between the underlying and the surface morphological structure. We further experiment with automated models for morpheme segmentation, where we find allomorphy as the major reason for segmentation errors. Finally, we show that subword tokenization algorithms are not viable for discovering morphemes in low-resource scenarios.
comment: Submitted to the 7th Workshop on Research in Computational Linguistic Typology and Multilingual NLP (SIGTYP)
♻ ☆ Improving Semantic Understanding in Speech Language Models via Brain-tuning ICLR 2025
Speech language models align with human brain responses to natural language to an impressive degree. However, current models rely heavily on low-level speech features, indicating they lack brain-relevant semantics which limits their utility as model organisms of semantic processing in the brain. In this work, we address this limitation by inducing brain-relevant bias directly into the models via fine-tuning with fMRI recordings of people listening to natural stories, a process we name brain-tuning. After testing it on 3 different pretrained model families, we show that brain-tuning not only improves overall alignment with new brain recordings in semantic language regions, but also reduces the reliance on low-level speech features for this alignment. Excitingly, we further show that brain-tuning leads to 1) consistent improvements in performance on a range of downstream tasks and 2) a representational space with increased semantic preference. Our results provide converging evidence, for the first time, that incorporating brain signals into the training of language models improves the models' semantic understanding.
comment: Published as a conference paper at ICLR 2025
♻ ☆ ADePT: Adaptive Decomposed Prompt Tuning for Parameter-Efficient Fine-tuning
Prompt Tuning (PT) enables the adaptation of Pre-trained Large Language Models (PLMs) to downstream tasks by optimizing a small amount of soft virtual tokens, which are prepended to the input token embeddings. Recently, Decomposed Prompt Tuning (DePT) has demonstrated superior adaptation capabilities by decomposing the soft prompt into a shorter soft prompt and a pair of low-rank matrices. The product of the pair of low-rank matrices is added to the input token embeddings to offset them. Additionally, DePT achieves faster inference compared to PT due to the shorter soft prompt. However, in this paper, we find that the position-based token embedding offsets of DePT restrict its ability to generalize across diverse model inputs, and that the shared embedding offsets across many token embeddings result in sub-optimization. To tackle these issues, we introduce Adaptive Decomposed Prompt Tuning (ADePT), which is composed of a short soft prompt and a shallow token-shared feed-forward neural network. ADePT utilizes the token-shared feed-forward neural network to learn the embedding offsets for each token, enabling adaptive embedding offsets that vary according to the model input and better optimization of token embedding offsets. This enables ADePT to achieve superior adaptation performance without requiring more inference time or additional trainable parameters compared to vanilla PT and its variants. In comprehensive experiments across 23 natural language processing tasks and 4 typical PLMs of different scales, ADePT consistently surpasses the other leading parameter-efficient fine-tuning methods, and even outperforms the full fine-tuning in certain scenarios. We also provide a theoretical analysis towards ADePT. Code is available at https://github.com/HungerPWAY/ADePT.
♻ ☆ EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty
Autoregressive decoding makes the inference of Large Language Models (LLMs) time-consuming. In this paper, we reconsider speculative sampling and derive two key observations. Firstly, autoregression at the feature (second-to-top-layer) level is more straightforward than at the token level. Secondly, the inherent uncertainty in feature (second-to-top-layer) level autoregression constrains its performance. Based on these insights, we introduce EAGLE (Extrapolation Algorithm for Greater Language-model Efficiency), a simple yet highly efficient speculative sampling framework. By incorporating a token sequence advanced by one time step, EAGLE effectively resolves the uncertainty, enabling precise second-to-top-layer feature prediction with minimal overhead. We conducted comprehensive evaluations of EAGLE, including all models from the Vicuna and LLaMA2-Chat series, the MoE model Mixtral 8x7B Instruct, and tasks in dialogue, code generation, mathematical reasoning, and instruction following. For LLaMA2-Chat 70B, EAGLE achieved a latency speedup ratio of 2.7x-3.5x, doubled throughput, while maintaining the distribution of the generated text.
♻ ☆ Do we still need Human Annotators? Prompting Large Language Models for Aspect Sentiment Quad Prediction
Aspect sentiment quadruple prediction (ASQP) facilitates a detailed understanding of opinions expressed in a text by identifying the opinion term, aspect term, aspect category and sentiment polarity for each opinion. However, annotating a full set of training examples to fine-tune models for ASQP is a resource-intensive process. In this study, we explore the capabilities of large language models (LLMs) for zero- and few-shot learning on the ASQP task across five diverse datasets. We report F1 scores slightly below those obtained with state-of-the-art fine-tuned models but exceeding previously reported zero- and few-shot performance. In the 40-shot setting on the Rest16 restaurant domain dataset, LLMs achieved an F1 score of 52.46, compared to 60.39 by the best-performing fine-tuned method MVP. Additionally, we report the performance of LLMs in target aspect sentiment detection (TASD), where the F1 scores were also close to fine-tuned models, achieving 66.03 on Rest16 in the 40-shot setting, compared to 72.76 with MVP. While human annotators remain essential for achieving optimal performance, LLMs can reduce the need for extensive manual annotation in ASQP tasks.
♻ ☆ PAD: Personalized Alignment of LLMs at Decoding-Time ICLR 2025
Aligning with personalized preferences, which vary significantly across cultural, educational, and political differences, poses a significant challenge due to the computational costs and data demands of traditional alignment methods. In response, this paper presents Personalized Alignment at Decoding-time (PAD), a novel framework designed to align LLM outputs with diverse personalized preferences during the inference phase, eliminating the need for additional training. By introducing a unique personalized reward modeling strategy, this framework decouples the text generation process from personalized preferences, facilitating the generation of generalizable token-level personalized rewards. The PAD algorithm leverages these rewards to guide the decoding process, dynamically tailoring the base model's predictions to personalized preferences. Extensive experimental results demonstrate that PAD not only outperforms existing training-based alignment methods in terms of aligning with diverse preferences but also shows significant generalizability to preferences unseen during training and scalability across different base models. This work advances the capability of LLMs to meet user needs in real-time applications, presenting a substantial step forward in personalized LLM alignment.
comment: ICLR 2025
♻ ☆ Reap the Wild Wind: Detecting Media Storms in Large-Scale News Corpora EMNLP 2024
Media Storms, dramatic outbursts of attention to a story, are central components of media dynamics and the attention landscape. Despite their significance, there has been little systematic and empirical research on this concept due to issues of measurement and operationalization. We introduce an iterative human-in-the-loop method to identify media storms in a large-scale corpus of news articles. The text is first transformed into signals of dispersion based on several textual characteristics. In each iteration, we apply unsupervised anomaly detection to these signals; each anomaly is then validated by an expert to confirm the presence of a storm, and those results are then used to tune the anomaly detection in the next iteration. We demonstrate the applicability of this method in two scenarios: first, supplementing an initial list of media storms within a specific time frame; and second, detecting media storms in new time periods. We make available a media storm dataset compiled using both scenarios. Both the method and dataset offer the basis for comprehensive empirical research into the concept of media storms, including characterizing them and predicting their outbursts and durations, in mainstream media or social media platforms.
comment: This paper was accepted and published in Findings of EMNLP 2024. The final version is available at: https://aclanthology.org/2024.findings-emnlp.275/
♻ ☆ Let the Code LLM Edit Itself When You Edit the Code ICLR 2025
In this work, we investigate a typical scenario in code generation where a developer edits existing code in real time and requests a code assistant, e.g., a large language model, to re-predict the next token or next line on the fly. Naively, the LLM needs to re-encode the entire KV cache to provide an accurate prediction. However, this process is computationally expensive, especially when the sequence length is long. Simply encoding the edited subsequence and integrating it to the original KV cache meets the temporal confusion problem, leading to significantly worse performance. We address this efficiency and accuracy trade-off by introducing \underline{\textbf{Positional \textbf{I}ntegrity \textbf{E}ncoding} (PIE). Building upon the rotary positional encoding, PIE first removes the rotary matrices in the Key cache that introduce temporal confusion and then reapplies the correct rotary matrices. This process ensures that positional relationships between tokens are correct and requires only a single round of matrix multiplication. We validate the effectiveness of PIE through extensive experiments on the RepoBench-C-8k dataset, utilizing DeepSeek-Coder models with 1.3B, 6.7B, and 33B parameters. Our evaluation includes three real-world coding tasks: code insertion, code deletion, and multi-place code editing. Results demonstrate that PIE reduces computational overhead by over 85% compared to the standard full recomputation approach across all model sizes and tasks while well approximating the model performance.
comment: ICLR 2025 Camera Ready
♻ ☆ DOVE: A Large-Scale Multi-Dimensional Predictions Dataset Towards Meaningful LLM Evaluation
Recent work found that LLMs are sensitive to a wide range of arbitrary prompt dimensions, including the type of delimiters, answer enumerators, instruction wording, and more. This throws into question popular single-prompt evaluation practices. We present DOVE (Dataset Of Variation Evaluation) a large-scale dataset containing prompt perturbations of various evaluation benchmarks. In contrast to previous work, we examine LLM sensitivity from an holistic perspective, and assess the joint effects of perturbations along various dimensions, resulting in thousands of perturbations per instance. We evaluate several model families against DOVE, leading to several findings, including efficient methods for choosing well-performing prompts, observing that few-shot examples reduce sensitivity, and identifying instances which are inherently hard across all perturbations. DOVE consists of more than 250M prompt perturbations and model outputs, which we make publicly available to spur a community-wide effort toward meaningful, robust, and efficient evaluation. Browse the data, contribute, and more: https://slab-nlp.github.io/DOVE/
♻ ☆ DermaSynth: Rich Synthetic Image-Text Pairs Using Open Access Dermatology Datasets
A major barrier to developing vision large language models (LLMs) in dermatology is the lack of large image--text pairs dataset. We introduce DermaSynth, a dataset comprising of 92,020 synthetic image--text pairs curated from 45,205 images (13,568 clinical and 35,561 dermatoscopic) for dermatology-related clinical tasks. Leveraging state-of-the-art LLMs, using Gemini 2.0, we used clinically related prompts and self-instruct method to generate diverse and rich synthetic texts. Metadata of the datasets were incorporated into the input prompts by targeting to reduce potential hallucinations. The resulting dataset builds upon open access dermatological image repositories (DERM12345, BCN20000, PAD-UFES-20, SCIN, and HIBA) that have permissive CC-BY-4.0 licenses. We also fine-tuned a preliminary Llama-3.2-11B-Vision-Instruct model, DermatoLlama 1.0, on 5,000 samples. We anticipate this dataset to support and accelerate AI research in dermatology. Data and code underlying this work are accessible at https://github.com/abdurrahimyilmaz/DermaSynth.
comment: 12 pages, 4 figures
♻ ☆ LLaSE-G1: Incentivizing Generalization Capability for LLaMA-based Speech Enhancement
Recent advancements in language models (LMs) have demonstrated strong capabilities in semantic understanding and contextual modeling, which have flourished in generative speech enhancement (SE). However, many LM-based SE approaches primarily focus on semantic information, often neglecting the critical role of acoustic information, which leads to acoustic inconsistency after enhancement and limited generalization across diverse SE tasks. In this paper, we introduce LLaSE-G1, a LLaMA-based language model that incentivizes generalization capabilities for speech enhancement. LLaSE-G1 offers the following key contributions: First, to mitigate acoustic inconsistency, LLaSE-G1 employs continuous representations from WavLM as input and predicts speech tokens from X-Codec2, maximizing acoustic preservation. Second, to promote generalization capability, LLaSE-G1 introduces dual-channel inputs and outputs, unifying multiple SE tasks without requiring task-specific IDs. Third, LLaSE-G1 outperforms prior task-specific discriminative and generative SE models, demonstrating scaling effects at test time and emerging capabilities for unseen SE tasks. Additionally, we release our code and models to support further research in this area.
comment: 13 pages, 2 figures, 8 tables
♻ ☆ Smoothing Grounding and Reasoning for MLLM-Powered GUI Agents with Query-Oriented Pivot Tasks
Perception-enhanced pre-training, particularly through grounding techniques, is widely adopted to enhance the performance of graphical user interface (GUI) agents. However, in resource-constrained scenarios, the format discrepancy between coordinate-oriented grounding and action-oriented reasoning limits the effectiveness of grounding for reasoning tasks. To address this challenge, we propose a query-oriented pivot approach called query inference, which serves as a bridge between GUI grounding and reasoning. By inferring potential user queries from a screenshot and its associated element coordinates, query inference improves the understanding of coordinates while aligning more closely with reasoning tasks. Experimental results show that query inference outperforms previous grounding techniques under the same training data scale. Notably, query inference achieves comparable or even better performance to large-scale grounding-enhanced OS-Atlas with less than 0.1% of training data. Furthermore, we explore the impact of reasoning formats and demonstrate that integrating additional semantic information into the input further boosts reasoning performance. The code is publicly available at https://github.com/ZrW00/GUIPivot.
♻ ☆ Tool Learning in the Wild: Empowering Language Models as Automatic Tool Agents WWW 2025
Augmenting large language models (LLMs) with external tools has emerged as a promising approach to extend their utility, enabling them to solve practical tasks. Previous methods manually parse tool documentation and create in-context demonstrations, transforming tools into structured formats for LLMs to use in their step-by-step reasoning. However, this manual process requires domain expertise and struggles to scale to large toolsets. Additionally, these methods rely heavily on ad-hoc inference techniques or special tokens to integrate free-form LLM generation with tool-calling actions, limiting the LLM's flexibility in handling diverse tool specifications and integrating multiple tools. In this work, we propose AutoTools, a framework that enables LLMs to automate the tool-use workflow. Specifically, the LLM automatically transforms tool documentation into callable functions, verifying syntax and runtime correctness. Then, the LLM integrates these functions into executable programs to solve practical tasks, flexibly grounding tool-use actions into its reasoning processes. Extensive experiments on existing and newly collected, more challenging benchmarks illustrate the superiority of our framework. Inspired by these promising results, we further investigate how to improve the expertise of LLMs, especially open-source LLMs with fewer parameters, within AutoTools. Thus, we propose the AutoTools-learning approach, training the LLMs with three learning tasks on 34k instances of high-quality synthetic data, including documentation understanding, relevance learning, and function programming. Fine-grained results validate the effectiveness of our overall training approach and each individual task. Our methods are an important step towards the use of LLMs for solving real-world tasks with external tools.
comment: Accepted by WWW 2025
♻ ☆ How Ambiguous Are the Rationales for Natural Language Reasoning? A Simple Approach to Handling Rationale Uncertainty COLING2025
The quality of rationales is essential in the reasoning capabilities of language models. Rationales not only enhance reasoning performance in complex natural language tasks but also justify model decisions. However, obtaining impeccable rationales is often impossible. Our study aims to investigate how ambiguous rationales play in model performances of natural language reasoning. We first assess the ambiguity of rationales through the lens of entropy and uncertainty in model prior beliefs, exploring its impact on task performance. We then propose a simple way to guide models to choose between two different reasoning paths depending on the ambiguity of the rationales. Our empirical results demonstrate that this approach leads to robust performance, particularly in adversarial scenarios where rationale quality is inconsistent.
comment: Accepted to COLING2025
♻ ☆ Evaluating Creative Short Story Generation in Humans and Large Language Models
Story-writing is a fundamental aspect of human imagination, relying heavily on creativity to produce narratives that are novel, effective, and surprising. While large language models (LLMs) have demonstrated the ability to generate high-quality stories, their creative story-writing capabilities remain under-explored. In this work, we conduct a systematic analysis of creativity in short story generation across 60 LLMs and 60 people using a five-sentence creative story-writing task. We use measures to automatically evaluate model- and human-generated stories across several dimensions of creativity, including novelty, surprise, diversity, and linguistic complexity. We also collect creativity ratings and Turing Test classifications from non-expert and expert human raters and LLMs. Automated metrics show that LLMs generate stylistically complex stories, but tend to fall short in terms of novelty, surprise and diversity when compared to average human writers. Expert ratings generally coincide with automated metrics. However, LLMs and non-experts rate LLM stories to be more creative than human-generated stories. We discuss why and how these differences in ratings occur, and their implications for both human and artificial creativity.
comment: Submitted to ICCC 2025
♻ ☆ From Matching to Generation: A Survey on Generative Information Retrieval
Information Retrieval (IR) systems are crucial tools for users to access information, which have long been dominated by traditional methods relying on similarity matching. With the advancement of pre-trained language models, generative information retrieval (GenIR) emerges as a novel paradigm, attracting increasing attention. Based on the form of information provided to users, current research in GenIR can be categorized into two aspects: \textbf{(1) Generative Document Retrieval} (GR) leverages the generative model's parameters for memorizing documents, enabling retrieval by directly generating relevant document identifiers without explicit indexing. \textbf{(2) Reliable Response Generation} employs language models to directly generate information users seek, breaking the limitations of traditional IR in terms of document granularity and relevance matching while offering flexibility, efficiency, and creativity to meet practical needs. This paper aims to systematically review the latest research progress in GenIR. We will summarize the advancements in GR regarding model training and structure, document identifier, incremental learning, etc., as well as progress in reliable response generation in aspects of internal knowledge memorization, external knowledge augmentation, etc. We also review the evaluation, challenges and future developments in GenIR systems. This review aims to offer a comprehensive reference for researchers, encouraging further development in the GenIR field. Github Repository: https://github.com/RUC-NLPIR/GenIR-Survey
♻ ☆ Multimodal Inconsistency Reasoning (MMIR): A New Benchmark for Multimodal Reasoning Models
Existing Multimodal Large Language Models (MLLMs) are predominantly trained and tested on consistent visual-textual inputs, leaving open the question of whether they can handle inconsistencies in real-world, layout-rich content. To bridge this gap, we propose the Multimodal Inconsistency Reasoning (MMIR) benchmark to assess MLLMs' ability to detect and reason about semantic mismatches in artifacts such as webpages, presentation slides, and posters. MMIR comprises 534 challenging samples, each containing synthetically injected errors across five reasoning-heavy categories: Factual Contradiction, Identity Misattribution, Contextual Mismatch, Quantitative Discrepancy, and Temporal/Spatial Incoherence. We evaluate six state-of-the-art MLLMs, showing that models with dedicated multimodal reasoning capabilities, such as o1, substantially outperform their counterparts while open-source models remain particularly vulnerable to inconsistency errors. Detailed error analyses further show that models excel in detecting pairwise inconsistencies but struggle with inconsistencies confined to single elements in complex layouts. Probing experiments reveal that single-modality prompting, including Chain-of-Thought (CoT) and Set-of-Mark (SoM) methods, yields marginal gains, revealing a key bottleneck in cross-modal reasoning. Our findings highlight the need for advanced multimodal reasoning and point to future research on multimodal inconsistency.
♻ ☆ Llamarine: Open-source Maritime Industry-specific Large Language Model
Large Language Models (LLMs) have demonstrated substantial potential in addressing complex reasoning tasks, yet their general-purpose nature often limits their effectiveness in specialized domains such as maritime navigation. To bridge this gap, we introduce Llamarine, the first open-source LLM designed specifically for maritime navigation. Llamarine 1.0 is developed through continued pretraining and fine-tuning on a high-quality corpus comprising maritime textbooks, research publications, and web text from Wikipedia. This domain-specific training enables the model to acquire expert-level knowledge in navigational principles, collision avoidance, route optimization, and regulatory compliance. Our key contributions include (a) the curation of a comprehensive maritime dataset from authoritative sources, ensuring depth and reliability in the model's knowledge base; (b) the development of a foundational model capable of reasoning about complex navigational challenges with greater accuracy than general-purpose LLMs; and (c) the establishment of a benchmark to evaluate performance in maritime-specific decision-making tasks. Experimental results demonstrate that Llamarine outperforms both general-purpose and commercial LLMs in critical navigation-related tasks, such as trajectory planning, risk assessment, and compliance with maritime regulations. By providing an open-source foundation model trained exclusively on high-quality maritime literature, Llamarine paves the way for AI-driven advancements in maritime safety, efficiency, and operational decision-making.
comment: Work in progress
♻ ☆ M2Lingual: Enhancing Multilingual, Multi-Turn Instruction Alignment in Large Language Models
Instruction finetuning (IFT) is critical for aligning Large Language Models (LLMs) to follow instructions. While many effective IFT datasets have been introduced recently, they predominantly focus on high-resource languages like English. To better align LLMs across a broad spectrum of languages and tasks, we propose a fully synthetic, novel taxonomy (Evol) guided Multilingual, Multi-turn instruction finetuning dataset, called M2Lingual. It is constructed by first selecting a diverse set of seed examples and then utilizing the proposed Evol taxonomy to convert these seeds into complex and challenging multi-turn instructions. We demonstrate the effectiveness of M2Lingual by training LLMs of varying sizes and showcasing the enhanced performance across a diverse set of languages. We contribute the 2 step Evol taxonomy with the guided generation code: https://github.com/ServiceNow/M2Lingual, as well as the first fully synthetic, general and task-oriented, multi-turn, multilingual dataset built with Evol - M2Lingual: https://huggingface.co/datasets/ServiceNow-AI/ M2Lingual - containing 182K total IFT pairs, covering 70 languages and 17+ NLP tasks.
comment: 39 pages
♻ ☆ SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction ICLR 2025
Large Language Models (LLMs) have demonstrated improved generation performance by incorporating externally retrieved knowledge, a process known as retrieval-augmented generation (RAG). Despite the potential of this approach, existing studies evaluate RAG effectiveness by 1) assessing retrieval and generation components jointly, which obscures retrieval's distinct contribution, or 2) examining retrievers using traditional metrics such as NDCG, which creates a gap in understanding retrieval's true utility in the overall generation process. To address the above limitations, in this work, we introduce an automatic evaluation method that measures retrieval quality through the lens of information gain within the RAG framework. Specifically, we propose Semantic Perplexity (SePer), a metric that captures the LLM's internal belief about the correctness of the retrieved information. We quantify the utility of retrieval by the extent to which it reduces semantic perplexity post-retrieval. Extensive experiments demonstrate that SePer not only aligns closely with human preferences but also offers a more precise and efficient evaluation of retrieval utility across diverse RAG scenarios.
comment: ICLR 2025 Spotlight
♻ ☆ Diversifying Question Generation over Knowledge Base via External Natural Questions
Previous methods on knowledge base question generation (KBQG) primarily focus on enhancing the quality of a single generated question. Recognizing the remarkable paraphrasing ability of humans, we contend that diverse texts should convey the same semantics through varied expressions. The above insights make diversifying question generation an intriguing task, where the first challenge is evaluation metrics for diversity. Current metrics inadequately assess the above diversity since they calculate the ratio of unique n-grams in the generated question itself, which leans more towards measuring duplication rather than true diversity. Accordingly, we devise a new diversity evaluation metric, which measures the diversity among top-k generated questions for each instance while ensuring their relevance to the ground truth. Clearly, the second challenge is how to enhance diversifying question generation. To address this challenge, we introduce a dual model framework interwoven by two selection strategies to generate diverse questions leveraging external natural questions. The main idea of our dual framework is to extract more diverse expressions and integrate them into the generation model to enhance diversifying question generation. Extensive experiments on widely used benchmarks for KBQG demonstrate that our proposed approach generates highly diverse questions and improves the performance of question answering tasks.
comment: 13 pages, 2 figures
♻ ☆ VideoRAG: Retrieval-Augmented Generation over Video Corpus
Retrieval-Augmented Generation (RAG) is a powerful strategy for improving the factual accuracy of models by retrieving external knowledge relevant to queries and incorporating it into the generation process. However, existing approaches primarily focus on text, with some recent advancements considering images, and they largely overlook videos, a rich source of multimodal knowledge capable of representing contextual details more effectively than any other modality. While very recent studies explore the use of videos in response generation, they either predefine query-associated videos without retrieval or convert videos into textual descriptions losing multimodal richness. To tackle these, we introduce VideoRAG, a framework that not only dynamically retrieves videos based on their relevance with queries but also utilizes both visual and textual information. The operation of VideoRAG is powered by recent Large Video Language Models (LVLMs), which enable the direct processing of video content to represent it for retrieval and the seamless integration of retrieved videos jointly with queries for response generation. Also, inspired by that the context size of LVLMs may not be sufficient to process all frames in extremely long videos and not all frames are equally important, we introduce a video frame selection mechanism to extract the most informative subset of frames, along with a strategy to extract textual information from videos (as it can aid the understanding of video content) when their subtitles are not available. We experimentally validate the effectiveness of VideoRAG, showcasing that it is superior to relevant baselines. Code is available at https://github.com/starsuzi/VideoRAG.
♻ ☆ Danoliteracy of Generative Large Language Models
The language technology moonshot moment of Generative Large Language Models (GLLMs) was not limited to English: These models brought a surge of technological applications, investments, and hype to low-resource languages as well. However, the capabilities of these models in languages such as Danish were, until recently, difficult to verify beyond qualitative demonstrations due to a lack of applicable evaluation corpora. We present a GLLM benchmark to evaluate \emph{Danoliteracy}, a measure of Danish language and cultural competency across eight diverse scenarios such as Danish citizenship tests and abstractive social media question answering. This limited-size benchmark was found to produce a robust ranking that correlates to human feedback at $\rho \sim 0.8$ with GPT-4 and Claude Opus models achieving the highest rankings. Analyzing these model results across scenarios, we find one strong underlying factor explaining $95\%$ of scenario performance variance for GLLMs in Danish, suggesting a $g$ factor of model consistency in language adaptation.
comment: 16 pages, 13 figures, Accepted to NoDaLiDa/Baltic-HLT 2025
♻ ☆ Beyond Exact Match: Semantically Reassessing Event Extraction by Large Language Models
Event extraction has gained extensive research attention due to its broad range of applications. However, the current mainstream evaluation method for event extraction relies on token-level exact match, which misjudges numerous semantic-level correct cases. This reliance leads to a significant discrepancy between the evaluated performance of models under exact match criteria and their real performance. To address this problem, we propose a reliable and semantic evaluation framework for event extraction, named RAEE, which accurately assesses extraction results at semantic-level instead of token-level. Specifically, RAEE leverages large language models (LLMs) as evaluation agents, incorporating an adaptive mechanism to achieve adaptive evaluations for precision and recall of triggers and arguments. Extensive experiments demonstrate that: (1) RAEE achieves a very strong correlation with human judgments; (2) after reassessing 14 models, including advanced LLMs, on 10 datasets, there is a significant performance gap between exact match and RAEE. The exact match evaluation significantly underestimates the performance of existing event extraction models, and in particular underestimates the capabilities of LLMs; (3) fine-grained analysis under RAEE evaluation reveals insightful phenomena worth further exploration. The evaluation toolkit of our proposed RAEE is publicly released.
♻ ☆ Crosslingual Capabilities and Knowledge Barriers in Multilingual Large Language Models
Large language models (LLMs) are typically multilingual due to pretraining on diverse multilingual corpora. But can these models relate corresponding concepts across languages, i.e., be crosslingual? This study evaluates state-of-the-art LLMs on inherently crosslingual tasks. We observe that while these models show promising surface-level crosslingual abilities on machine translation and embedding space analyses, they struggle with deeper crosslingual knowledge transfer, revealing a crosslingual knowledge barrier in both general (MMLU benchmark) and domain-specific (Harry Potter quiz and TOFU benchmark) contexts. Since simple inference-time mitigation methods offer only limited improvement, we propose fine-tuning of LLMs on mixed-language data, which effectively reduces these gaps, even when using out-of-domain datasets like WikiText. Our findings suggest the need for explicit optimization to unlock the full crosslingual potential of LLMs. Our code is publicly available at https://github.com/google-research/crosslingual-knowledge-barriers.
♻ ☆ SECURA: Sigmoid-Enhanced CUR Decomposition with Uninterrupted Retention and Low-Rank Adaptation in Large Language Models
With the rapid development of large language models (LLMs), fully fine-tuning (FT) these models is becoming increasingly infeasible due to high computational demands. Moreover, FT also increases the risk of catastrophic forgetting. As an alternative, Low-Rank Adaptation (LoRA) has been proposed. By fine-tuning only a small subset of parameters, LoRA achieves performance similar to FT while significantly reducing resource requirements. However, since LoRA inherits FT's design, the issue of catastrophic forgetting still remains. To address these limitations, we propose SECURA: Sigmoid-Enhanced CUR Decomposition LoRA, a novel PEFT variant designed to mitigate catastrophic forgetting while improving fine-tuning performance. Our method introduces a novel normalization technique, Sigmoid-based Magnitude Norm (S-MagNorm), which enhances parameter retention and fine-tuning efficiency. SECURA has been evaluated on a diverse range of tasks, including mathematical problem-solving (GSM8K), complex question-answering (CNNDM), translation (NewsDE), and complex multiple-choice reasoning (LogiQA). Experimental results demonstrate that it achieves an average fine-tuning improvement of 3.59% across four MCQ tasks and 2.51% across five QA tasks on Gemma2 2B, Qwen2 1.5B, Qwen2 7B, Llama3 8B, and Llama3.1 8B, outperforming DoRA. Additionally, SECURA demonstrates superior knowledge retention capabilities, achieving state-of-the-art performance in 16 continual learning tests and maintaining more than 70% accuracy on LLMs' basic knowledge compared to Experience Replay (ER), sequential learning (SEQ), EWC, I-LoRA, and CUR-LoRA.
comment: New work on PEFT for LLMs, introducing S-MagNorm and CABR-LoRA to enhance fine-tuning performance and knowledge retention. In v4, we renamed Sigmoid-based Magnitude Normalization to S-MagNorm for clarity and added a gradient comparison between SECURA and CABR-LoRA to highlight their contributions
♻ ☆ NitiBench: A Comprehensive Studies of LLM Frameworks Capabilities for Thai Legal Question Answering
The application of large language models (LLMs) in the legal domain holds significant potential for information retrieval and question answering, yet Thai legal QA systems face challenges due to a lack of standardized evaluation benchmarks and the complexity of Thai legal structures. This paper introduces NitiBench, a benchmark comprising two datasets: the NitiBench-CCL, covering general Thai financial law, and the NitiBench-Tax, which includes real-world tax law cases requiring advanced legal reasoning. We evaluate retrieval-augmented generation (RAG) and long-context LLM-based approaches to address three key research questions: the impact of domain-specific components like section-based chunking and cross-referencing, the comparative performance of different retrievers and LLMs, and the viability of long-context LLMs as an alternative to RAG. Our results show that section-based chunking significantly improves retrieval and end-to-end performance, current retrievers struggle with complex queries, and long-context LLMs still underperform RAG-based systems in Thai legal QA. To support fair evaluation, we propose tailored multi-label retrieval metrics and the use of an LLM-as-judge for coverage and contradiction detection method. These findings highlight the limitations of current Thai legal NLP solutions and provide a foundation for future research in the field. We also open-sourced our codes and dataset to available publicly.
♻ ☆ B-STaR: Monitoring and Balancing Exploration and Exploitation in Self-Taught Reasoners ICLR 2025
In the absence of extensive human-annotated data for complex reasoning tasks, self-improvement -- where models are trained on their own outputs -- has emerged as a primary method for enhancing performance. However, the critical factors underlying the mechanism of these iterative self-improving methods remain poorly understood, such as under what conditions self-improvement is effective, and what are the bottlenecks in the current iterations. In this work, we identify and propose methods to monitor two pivotal factors in this iterative process: (1) the model's ability to generate sufficiently diverse responses (exploration); and (2) the effectiveness of external rewards in distinguishing high-quality candidates from lower-quality ones (exploitation). Using mathematical reasoning as a case study, we begin with a quantitative analysis to track the dynamics of exploration and exploitation, discovering that a model's exploratory capabilities rapidly deteriorate over iterations, and the effectiveness of exploiting external rewards diminishes as well. Motivated by these findings, we introduce B-STaR, a Self-Taught Reasoning framework that autonomously adjusts configurations across iterations to Balance exploration and exploitation, thereby optimizing the self-improving effectiveness based on the current policy model and available rewards. Our experiments on mathematical reasoning, coding, and commonsense reasoning demonstrate that B-STaR not only enhances the model's exploratory capabilities throughout training but also achieves a more effective balance between exploration and exploitation, leading to superior performance.
comment: Published as a conference paper at ICLR 2025
♻ ☆ Frame-Voyager: Learning to Query Frames for Video Large Language Models ICLR 2025
Video Large Language Models (Video-LLMs) have made remarkable progress in video understanding tasks. However, they are constrained by the maximum length of input tokens, making it impractical to input entire videos. Existing frame selection approaches, such as uniform frame sampling and text-frame retrieval, fail to account for the information density variations in the videos or the complex instructions in the tasks, leading to sub-optimal performance. In this paper, we propose Frame-Voyager that learns to query informative frame combinations, based on the given textual queries in the task. To train Frame-Voyager, we introduce a new data collection and labeling pipeline, by ranking frame combinations using a pre-trained Video-LLM. Given a video of M frames, we traverse its T-frame combinations, feed them into a Video-LLM, and rank them based on Video-LLM's prediction losses. Using this ranking as supervision, we train Frame-Voyager to query the frame combinations with lower losses. In experiments, we evaluate Frame-Voyager on four Video Question Answering benchmarks by plugging it into two different Video-LLMs. The experimental results demonstrate that Frame-Voyager achieves impressive results in all settings, highlighting its potential as a plug-and-play solution for Video-LLMs.
comment: ICLR 2025, Camera-ready Version
♻ ☆ Detecting LLM-Generated Korean Text through Linguistic Feature Analysis
The rapid advancement of large language models (LLMs) increases the difficulty of distinguishing between human-written and LLM-generated text. Detecting LLM-generated text is crucial for upholding academic integrity, preventing plagiarism, protecting copyrights, and ensuring ethical research practices. Most prior studies on detecting LLM-generated text focus primarily on English text. However, languages with distinct morphological and syntactic characteristics require specialized detection approaches. Their unique structures and usage patterns can hinder the direct application of methods primarily designed for English. Among such languages, we focus on Korean, which has relatively flexible spacing rules, a rich morphological system, and less frequent comma usage compared to English. We introduce KatFish, the first benchmark dataset for detecting LLM-generated Korean text. The dataset consists of text written by humans and generated by four LLMs across three genres. By examining spacing patterns, part-of-speech diversity, and comma usage, we illuminate the linguistic differences between human-written and LLM-generated Korean text. Building on these observations, we propose KatFishNet, a detection method specifically designed for the Korean language. KatFishNet achieves an average of 19.78% higher AUROC compared to the best-performing existing detection method. Our code and data are available at https://github.com/Shinwoo-Park/detecting_llm_generated_korean_text_through_linguistic_analysis.
♻ ☆ On Memorization of Large Language Models in Logical Reasoning
Large language models (LLMs) achieve good performance on challenging reasoning benchmarks, yet could also make basic reasoning mistakes. This contrasting behavior is puzzling when it comes to understanding the mechanisms behind LLMs' reasoning capabilities. One hypothesis is that the increasingly high and nearly saturated performance on common reasoning benchmarks could be due to the memorization of similar problems. In this paper, we systematically investigate this hypothesis with a quantitative measurement of memorization in reasoning tasks, using a dynamically generated logical reasoning benchmark based on Knights and Knaves (K&K) puzzles. We find that LLMs could interpolate and memorize the training puzzles (achieving near-perfect accuracy) after fine-tuning, yet they struggle with slight variations of these puzzles. On the other hand, we show that while fine-tuning leads to heavy memorization, it also consistently improves generalization performance. Through in-depth analyses with perturbation tests, cross difficulty-level transferability, probing model internals, and fine-tuning with wrong answers, we establish that LLMs develop reasoning skills on K&K puzzles alongside memorization. Finally, our analysis based on a per-sample memorization score sheds light on how LLMs switch between reasoning and memorization when solving logical puzzles. Our code and data are available at https://memkklogic.github.io.
♻ ☆ Predictive Data Selection: The Data That Predicts Is the Data That Teaches
Language model pretraining involves training on extensive corpora, where data quality plays a pivotal role. In this work, we aim to directly estimate the contribution of data during pretraining and select pretraining data in an efficient manner. Specifically, we draw inspiration from recent findings showing that compression efficiency (i.e., the normalized loss) of diverse models on certain text correlates strongly with their downstream performance, when the text domain aligns with the downstream benchmarks(Huang et al., 2024). Building on this observation, we hypothesize that data on which model losses are predictive of downstream abilities also contribute effectively to learning. To leverage this insight, we introduce predictive data selection (PreSelect), a lightweight and efficient data selection method that requires training and deploying only a fastText-based scorer. Through comprehensive experiments with 1B and 3B parameter models, we demonstrate that models trained on 30B tokens selected with PreSelect surpass the performance of the vanilla baseline trained on 300B tokens, achieving a 10x reduction in compute requirements. Furthermore, PreSelect significantly outperforms other competitive data selection baselines, such as DCLM and FineWeb-Edu on a scale of 3B models trained on 100B tokens. We open-source our trained data selection scorer along with the curated datasets at https://github.com/hkust-nlp/PreSelect.
comment: 21 pages
♻ ☆ RAG-DDR: Optimizing Retrieval-Augmented Generation Using Differentiable Data Rewards
Retrieval-Augmented Generation (RAG) has proven its effectiveness in mitigating hallucinations in Large Language Models (LLMs) by retrieving knowledge from external resources. To adapt LLMs for the RAG systems, current approaches use instruction tuning to optimize LLMs, improving their ability to utilize retrieved knowledge. This supervised fine-tuning (SFT) approach focuses on equipping LLMs to handle diverse RAG tasks using different instructions. However, it trains RAG modules to overfit training signals and overlooks the varying data preferences among agents within the RAG system. In this paper, we propose a Differentiable Data Rewards (DDR) method, which end-to-end trains RAG systems by aligning data preferences between different RAG modules. DDR works by collecting the rewards to optimize each agent in the RAG system with the rollout method, which prompts agents to sample some potential responses as perturbations, evaluates the impact of these perturbations on the whole RAG system, and subsequently optimizes the agent to produce outputs that improve the performance of the RAG system. Our experiments on various knowledge-intensive tasks demonstrate that DDR significantly outperforms the SFT method, particularly for LLMs with smaller-scale parameters that depend more on the retrieved knowledge. Additionally, DDR exhibits a stronger capability to align the data preference between RAG modules. The DDR method makes the generation module more effective in extracting key information from documents and mitigating conflicts between parametric memory and external knowledge. All codes are available at https://github.com/OpenMatch/RAG-DDR.
♻ ☆ KiVA: Kid-inspired Visual Analogies for Testing Large Multimodal Models
This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.
comment: 10 pages. For the KiVA benchmark, see https://github.com/ey242/KiVA
♻ ☆ Tutorial on Using Machine Learning and Deep Learning Models for Mental Illness Detection
Social media has become an important source for understanding mental health, providing researchers with a way to detect conditions like depression from user-generated posts. This tutorial provides practical guidance to address common challenges in applying machine learning and deep learning methods for mental health detection on these platforms. It focuses on strategies for working with diverse datasets, improving text preprocessing, and addressing issues such as imbalanced data and model evaluation. Real-world examples and step-by-step instructions demonstrate how to apply these techniques effectively, with an emphasis on transparency, reproducibility, and ethical considerations. By sharing these approaches, this tutorial aims to help researchers build more reliable and widely applicable models for mental health research, contributing to better tools for early detection and intervention.
♻ ☆ Toxicity Detection towards Adaptability to Changing Perturbations
Toxicity detection is crucial for maintaining the peace of the society. While existing methods perform well on normal toxic contents or those generated by specific perturbation methods, they are vulnerable to evolving perturbation patterns. However, in real-world scenarios, malicious users tend to create new perturbation patterns for fooling the detectors. For example, some users may circumvent the detector of large language models (LLMs) by adding `I am a scientist' at the beginning of the prompt. In this paper, we introduce a novel problem, i.e., continual learning jailbreak perturbation patterns, into the toxicity detection field. To tackle this problem, we first construct a new dataset generated by 9 types of perturbation patterns, 7 of them are summarized from prior work and 2 of them are developed by us. We then systematically validate the vulnerability of current methods on this new perturbation pattern-aware dataset via both the zero-shot and fine tuned cross-pattern detection. Upon this, we present the domain incremental learning paradigm and the corresponding benchmark to ensure the detector's robustness to dynamically emerging types of perturbed toxic text. Our code and dataset are provided in the appendix and will be publicly available at GitHub, by which we wish to offer new research opportunities for the security-relevant communities.
comment: There are still some flaws in the uploaded content, which may cause confusion for readers. To be rigorous, we need to retract the paper for optimization and improvement
♻ ☆ Few-shot Personalization of LLMs with Mis-aligned Responses NAACL 25
As the diversity of users increases, the capability of providing personalized responses by large language models (LLMs) has become increasingly important. Existing approaches have only limited successes in LLM personalization, due to the absence of personalized learning or the reliance on shared personal data. This paper proposes a new approach for a few-shot personalization of LLMs with their mis-aligned responses (Fermi). Our key idea is to learn a set of personalized prompts for each user by progressively improving the prompts using LLMs, based on user profile (e.g., demographic information) and a few examples of previous opinions. During an iterative process of prompt improvement, we incorporate the contexts of mis-aligned responses by LLMs, which are especially crucial for the effective personalization of LLMs. In addition, we develop an effective inference method to further leverage the context of the test query and the personalized prompts. Our experimental results demonstrate that Fermi significantly improves performance across various benchmarks, compared to best-performing baselines.
comment: NAACL 25 (main, long), 32 pages
♻ ☆ Dataverse: Open-Source ETL (Extract, Transform, Load) Pipeline for Large Language Models NAACL 2025
To address the challenges associated with data processing at scale, we propose Dataverse, a unified open-source Extract-Transform-Load (ETL) pipeline for large language models (LLMs) with a user-friendly design at its core. Easy addition of custom processors with block-based interface in Dataverse allows users to readily and efficiently use Dataverse to build their own ETL pipeline. We hope that Dataverse will serve as a vital tool for LLM development and open source the entire library to welcome community contribution. Additionally, we provide a concise, two-minute video demonstration of our system, illustrating its capabilities and implementation.
comment: Accepted to NAACL 2025 Demo
♻ ☆ Evaluating Vision-Language Models as Evaluators in Path Planning
Despite their promise to perform complex reasoning, large language models (LLMs) have been shown to have limited effectiveness in end-to-end planning. This has inspired an intriguing question: if these models cannot plan well, can they still contribute to the planning framework as a helpful plan evaluator? In this work, we generalize this question to consider LLMs augmented with visual understanding, i.e., Vision-Language Models (VLMs). We introduce PathEval, a novel benchmark evaluating VLMs as plan evaluators in complex path-planning scenarios. Succeeding in the benchmark requires a VLM to be able to abstract traits of optimal paths from the scenario description, demonstrate precise low-level perception on each path, and integrate this information to decide the better path. Our analysis of state-of-the-art VLMs reveals that these models face significant challenges on the benchmark. We observe that the VLMs can precisely abstract given scenarios to identify the desired traits and exhibit mixed performance in integrating the provided information. Yet, their vision component presents a critical bottleneck, with models struggling to perceive low-level details about a path. Our experimental results show that this issue cannot be trivially addressed via end-to-end fine-tuning; rather, task-specific discriminative adaptation of these vision encoders is needed for these VLMs to become effective path evaluators.
♻ ☆ HaLoRA: Hardware-aware Low-Rank Adaptation for Large Language Models Based on Hybrid Compute-in-Memory Architecture
Low-rank adaptation (LoRA) is a predominant parameter-efficient finetuning method to adapt large language models (LLMs) for downstream tasks. In this paper, we first propose to deploy the LoRA-finetuned LLMs on the hybrid compute-in-memory (CIM) architecture (i.e., pretrained weights onto RRAM and LoRA onto SRAM). To address performance degradation from RRAM's inherent noise, we design a novel Hardware-aware Low-rank Adaption (HaLoRA) method, aiming to train a LoRA branch that is both robust and accurate by aligning the training objectives under both ideal and noisy conditions. Experiments finetuning LLaMA 3.2 1B and 3B demonstrate HaLoRA's effectiveness across multiple reasoning tasks, achieving up to 22.7 improvement in average score while maintaining robustness at various noise levels.
comment: 7 pages
♻ ☆ Modular Conversational Agents for Surveys and Interviews
Surveys and interviews are widely used for collecting insights on emerging or hypothetical scenarios. Traditional human-led methods often face challenges related to cost, scalability, and consistency. Recently, various domains have begun to explore the use of conversational agents (chatbots) powered by generative artificial intelligence (AI) technologies. However, considering decisions in transportation investments and policies often carry significant public and environmental stakes, surveys and interviews face unique challenges in integrating AI agents, underscoring the need for a rigorous, resource-efficient approach that enhances participant engagement and ensures privacy. This paper addresses this gap by introducing a modular approach and its resulting parameterized process for designing AI agents. We detail the system architecture, integrating engineered prompts, specialized knowledge bases, and customizable, goal-oriented conversational logic. We demonstrate the adaptability, generalizability, and efficacy of our modular approach through three empirical studies: (1) travel preference surveys, highlighting conditional logic and multimodal (voice, text, and image generation) capabilities; (2) public opinion elicitation on a newly constructed, novel infrastructure project, showcasing question customization and multilingual (English and French) capabilities; and (3) expert consultation about the impact of technologies on future transportation systems, highlighting real-time, clarification request capabilities for open-ended questions, resilience in handling erratic inputs, and efficient transcript postprocessing. The results suggest that the AI agent increases completion rates and response quality. Furthermore, the modular approach demonstrates controllability, flexibility, and robustness while addressing key ethical, privacy, security, and token consumption concerns.
♻ ☆ Confidential Prompting: Protecting User Prompts from Cloud LLM Providers
Our work tackles the challenge of securing user inputs in cloud-hosted large language model (LLM) serving while ensuring model confidentiality, output invariance, and compute efficiency. We introduce Secure Partitioned Decoding (SPD), which uses confidential computing to confine user prompts to a trusted execution environment (TEE), namely a confidential virtual machine (CVM), while allowing service providers to generate tokens efficiently. We also introduce a novel cryptographic method, Prompt Obfuscation (PO), to ensure robustness against reconstruction attacks on SPD. We demonstrate our approach preserves both prompt confidentiality and LLM serving efficiency. Our solution enables privacy-preserving cloud LLM serving that handles sensitive prompts, such as clinical records, financial data, and personal information.
♻ ☆ From Specific-MLLMs to Omni-MLLMs: A Survey on MLLMs Aligned with Multi-modalities
To tackle complex tasks in real-world scenarios, more researchers are focusing on Omni-MLLMs, which aim to achieve omni-modal understanding and generation. Beyond the constraints of any specific non-linguistic modality, Omni-MLLMs map various non-linguistic modalities into the embedding space of LLMs and enable the interaction and understanding of arbitrary combinations of modalities within a single model. In this paper, we systematically investigate relevant research and provide a comprehensive survey of Omni-MLLMs. Specifically, we first explain the four core components of Omni-MLLMs for unified multi-modal modeling with a meticulous taxonomy that offers novel perspectives. Then, we introduce the effective integration achieved through two-stage training and discuss the corresponding datasets as well as evaluation. Furthermore, we summarize the main challenges of current Omni-MLLMs and outline future directions. We hope this paper serves as an introduction for beginners and promotes the advancement of related research. Resources have been made publicly available at https://github.com/threegold116/Awesome-Omni-MLLMs.
comment: 35 pages
♻ ☆ Understanding LLM Development Through Longitudinal Study: Insights from the Open Ko-LLM Leaderboard NAACL 2025
This paper conducts a longitudinal study over eleven months to address the limitations of prior research on the Open Ko-LLM Leaderboard, which have relied on empirical studies with restricted observation periods of only five months. By extending the analysis duration, we aim to provide a more comprehensive understanding of the progression in developing Korean large language models (LLMs). Our study is guided by three primary research questions: (1) What are the specific challenges in improving LLM performance across diverse tasks on the Open Ko-LLM Leaderboard over time? (2) How does model size impact task performance correlations across various benchmarks? (3) How have the patterns in leaderboard rankings shifted over time on the Open Ko-LLM Leaderboard?. By analyzing 1,769 models over this period, our research offers a comprehensive examination of the ongoing advancements in LLMs and the evolving nature of evaluation frameworks.
comment: Accepted to NAACL 2025 Industry
♻ ☆ Open Ko-LLM Leaderboard2: Bridging Foundational and Practical Evaluation for Korean LLMs NAACL 2025
The Open Ko-LLM Leaderboard has been instrumental in benchmarking Korean Large Language Models (LLMs), yet it has certain limitations. Notably, the disconnect between quantitative improvements on the overly academic leaderboard benchmarks and the qualitative impact of the models should be addressed. Furthermore, the benchmark suite is largely composed of translated versions of their English counterparts, which may not fully capture the intricacies of the Korean language. To address these issues, we propose Open Ko-LLM Leaderboard2, an improved version of the earlier Open Ko-LLM Leaderboard. The original benchmarks are entirely replaced with new tasks that are more closely aligned with real-world capabilities. Additionally, four new native Korean benchmarks are introduced to better reflect the distinct characteristics of the Korean language. Through these refinements, Open Ko-LLM Leaderboard2 seeks to provide a more meaningful evaluation for advancing Korean LLMs.
comment: Accepted to NAACL 2025 Industry
♻ ☆ SARChat-Bench-2M: A Multi-Task Vision-Language Benchmark for SAR Image Interpretation
As a powerful all-weather Earth observation tool, synthetic aperture radar (SAR) remote sensing enables critical military reconnaissance, maritime surveillance, and infrastructure monitoring. Although Vision language models (VLMs) have made remarkable progress in natural language processing and image understanding, their applications remain limited in professional domains due to insufficient domain expertise. This paper innovatively proposes the first large-scale multimodal dialogue dataset for SAR images, named SARChat-2M, which contains approximately 2 million high-quality image-text pairs, encompasses diverse scenarios with detailed target annotations. This dataset not only supports several key tasks such as visual understanding and object detection tasks, but also has unique innovative aspects: this study develop a visual-language dataset and benchmark for the SAR domain, enabling and evaluating VLMs' capabilities in SAR image interpretation, which provides a paradigmatic framework for constructing multimodal datasets across various remote sensing vertical domains. Through experiments on 16 mainstream VLMs, the effectiveness of the dataset has been fully verified. The project will be released at https://github.com/JimmyMa99/SARChat.
♻ ☆ The Future Outcome Reasoning and Confidence Assessment Benchmark
Forecasting is an important task in many domains, such as technology and economics. However existing forecasting benchmarks largely lack comprehensive confidence assessment, focus on limited question types, and often consist of artificial questions that do not align with real-world human forecasting needs. To address these gaps, we introduce FOReCAst (Future Outcome Reasoning and Confidence Assessment), a benchmark that evaluates models' ability to make predictions and their confidence in them. FOReCAst spans diverse forecasting scenarios involving Boolean questions, timeframe prediction, and quantity estimation, enabling a comprehensive evaluation of both prediction accuracy and confidence calibration for real-world applications.
♻ ☆ LLaVA-Critic: Learning to Evaluate Multimodal Models CVPR 2025
We introduce LLaVA-Critic, the first open-source large multimodal model (LMM) designed as a generalist evaluator to assess performance across a wide range of multimodal tasks. LLaVA-Critic is trained using a high-quality critic instruction-following dataset that incorporates diverse evaluation criteria and scenarios. Our experiments demonstrate the model's effectiveness in two key areas: (1) LMM-as-a-Judge, where LLaVA-Critic provides reliable evaluation scores, performing on par with or surpassing GPT models on multiple evaluation benchmarks; and (2) Preference Learning, where it generates reward signals for preference learning, enhancing model alignment capabilities. This work underscores the potential of open-source LMMs in self-critique and evaluation, setting the stage for future research into scalable, superhuman alignment feedback mechanisms for LMMs.
comment: Accepted by CVPR 2025; Project Page: https://llava-vl.github.io/blog/2024-10-03-llava-critic
♻ ☆ CoRNStack: High-Quality Contrastive Data for Better Code Retrieval and Reranking ICLR 2025
Effective code retrieval plays a crucial role in advancing code generation, bug fixing, and software maintenance, particularly as software systems increase in complexity. While current code embedding models have demonstrated promise in retrieving code snippets for small-scale, well-defined tasks, they often underperform in more demanding real-world applications such as bug localization within GitHub repositories. We hypothesize that a key issue is their reliance on noisy and inconsistent datasets for training, which impedes their ability to generalize to more complex retrieval scenarios. To address these limitations, we introduce CoRNStack, a large-scale, high-quality contrastive training dataset for code that spans multiple programming languages. This dataset is curated using consistency filtering to eliminate noisy positives and is further enriched with mined hard negatives, thereby facilitating more effective learning. We demonstrate that contrastive training of embedding models using CoRNStack leads to state-of-the-art performance across a variety of code retrieval tasks. Furthermore, the dataset can be leveraged for training code reranking models, a largely underexplored area compared to text reranking. Our finetuned code reranking model significantly improves the ranking quality over the retrieved results. Finally, by employing our code retriever and reranker together, we demonstrate significant improvements in function localization for GitHub issues, an important component of real-world software development.
comment: Published as a conference paper at ICLR 2025. First and second author had equal contribution
♻ ☆ Measuring and Enhancing Trustworthiness of LLMs in RAG through Grounded Attributions and Learning to Refuse ICLR 2025
LLMs are an integral component of retrieval-augmented generation (RAG) systems. While many studies focus on evaluating the overall quality of end-to-end RAG systems, there is a gap in understanding the appropriateness of LLMs for the RAG task. To address this, we introduce Trust-Score, a holistic metric that evaluates the trustworthiness of LLMs within the RAG framework. Our results show that various prompting methods, such as in-context learning, fail to effectively adapt LLMs to the RAG task as measured by Trust-Score. Consequently, we propose Trust-Align, a method to align LLMs for improved Trust-Score performance. 26 out of 27 models aligned using Trust-Align substantially outperform competitive baselines on ASQA, QAMPARI, and ELI5. Specifically, in LLaMA-3-8b, Trust-Align outperforms FRONT on ASQA (up 12.56), QAMPARI (up 36.04), and ELI5 (up 17.69). Trust-Align also significantly enhances models' ability to correctly refuse and provide quality citations. We also demonstrate the effectiveness of Trust-Align across different open-weight models, including the LLaMA series (1b to 8b), Qwen-2.5 series (0.5b to 7b), and Phi3.5 (3.8b). We release our code at https://github.com/declare-lab/trust-align.
comment: Published at ICLR 2025 (Oral)
♻ ☆ miniCTX: Neural Theorem Proving with (Long-)Contexts
Real-world formal theorem proving often depends on a wealth of context, including definitions, lemmas, comments, file structure, and other information. We introduce miniCTX, which tests a model's ability to prove formal mathematical theorems that depend on new context that is not seen during training. miniCTX contains theorems sourced from real Lean projects and textbooks, each associated with a context that can span tens of thousands of tokens. Models are tasked with proving a theorem given access to code from the theorem's repository, which contains context that is needed for the proof. As a baseline for miniCTX, we tested fine-tuning and prompting methods that condition theorem proving on preceding context. Both approaches substantially outperform traditional methods that rely solely on state information. We found that this ability to use context is not captured by previous benchmarks such as miniF2F. Alongside miniCTX, we offer ntp-toolkit for automatically extracting and annotating theorem proving data, making it easy to add new projects into miniCTX to ensure that contexts are not seen during training. miniCTX offers a challenging and realistic evaluation of neural theorem provers.
comment: Project page: https://cmu-l3.github.io/minictx
♻ ☆ Spread Preference Annotation: Direct Preference Judgment for Efficient LLM Alignment ICLR 2025
Aligning large language models (LLMs) with human preferences becomes a key component to obtaining state-of-the-art performance, but it yields a huge cost to construct a large human-annotated preference dataset. To tackle this problem, we propose a new framework, Spread Preference Annotation with direct preference judgment (SPA), that boosts the alignment of LLMs using only a very small amount of human-annotated preference data. Our key idea is leveraging the human prior knowledge within the small (seed) data and progressively improving the alignment of LLM, by iteratively generating the responses and learning from them with the self-annotated preference data. To be specific, we propose to derive the preference label from the logits of LLM to explicitly extract the model's inherent preference. Compared to the previous approaches using external reward models or implicit in-context learning, we observe that the proposed approach is significantly more effective. In addition, we introduce a noise-aware preference learning algorithm to mitigate the risk of low quality within generated preference data. Our experimental results demonstrate that the proposed framework significantly boosts the alignment of LLMs. For example, we achieve superior alignment performance on AlpacaEval 2.0 with only 3.3% of the ground-truth preference labels in the Ultrafeedback data compared to the cases using the entire data or state-of-the-art baselines.
comment: ICLR 2025 Oral Presentation, 22 pages
♻ ☆ English Please: Evaluating Machine Translation for Multilingual Bug Reports
Accurate translation of bug reports is critical for efficient collaboration in global software development. In this study, we conduct the first comprehensive evaluation of machine translation (MT) performance on bug reports, analyzing the capabilities of DeepL, AWS Translate, and ChatGPT using data from the Visual Studio Code GitHub repository, specifically focusing on reports labeled with the english-please tag. To thoroughly assess the accuracy and effectiveness of each system, we employ multiple machine translation metrics, including BLEU, BERTScore, COMET, METEOR, and ROUGE. Our findings indicate that DeepL consistently outperforms the other systems across most automatic metrics, demonstrating strong lexical and semantic alignment. AWS Translate performs competitively, particularly in METEOR, while ChatGPT lags in key metrics. This study underscores the importance of domain adaptation for translating technical texts and offers guidance for integrating automated translation into bug-triaging workflows. Moreover, our results establish a foundation for future research to refine machine translation solutions for specialized engineering contexts. The code and dataset for this paper are available at GitHub: https://github.com/av9ash/gitbugs/tree/main/multilingual.
comment: 8 Pages, 4 Figures, 3 Tables
♻ ☆ LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory ICLR 2025
Recent large language model (LLM)-driven chat assistant systems have integrated memory components to track user-assistant chat histories, enabling more accurate and personalized responses. However, their long-term memory capabilities in sustained interactions remain underexplored. We introduce LongMemEval, a comprehensive benchmark designed to evaluate five core long-term memory abilities of chat assistants: information extraction, multi-session reasoning, temporal reasoning, knowledge updates, and abstention. With 500 meticulously curated questions embedded within freely scalable user-assistant chat histories, LongMemEval presents a significant challenge to existing long-term memory systems, with commercial chat assistants and long-context LLMs showing a 30% accuracy drop on memorizing information across sustained interactions. We then present a unified framework that breaks down the long-term memory design into three stages: indexing, retrieval, and reading. Built upon key experimental insights, we propose several memory design optimizations including session decomposition for value granularity, fact-augmented key expansion for indexing, and time-aware query expansion for refining the search scope. Extensive experiments show that these optimizations greatly improve both memory recall and downstream question answering on LongMemEval. Overall, our study provides valuable resources and guidance for advancing the long-term memory capabilities of LLM-based chat assistants, paving the way toward more personalized and reliable conversational AI. Our benchmark and code are publicly available at https://github.com/xiaowu0162/LongMemEval.
comment: ICLR 2025
♻ ☆ A Simplified Retriever to Improve Accuracy of Phenotype Normalizations by Large Language Models
Large language models (LLMs) have shown improved accuracy in phenotype term normalization tasks when augmented with retrievers that suggest candidate normalizations based on term definitions. In this work, we introduce a simplified retriever that enhances LLM accuracy by searching the Human Phenotype Ontology (HPO) for candidate matches using contextual word embeddings from BioBERT without the need for explicit term definitions. Testing this method on terms derived from the clinical synopses of Online Mendelian Inheritance in Man (OMIM), we demonstrate that the normalization accuracy of a state-of-the-art LLM increases from a baseline of 62.3% without augmentation to 90.3% with retriever augmentation. This approach is potentially generalizable to other biomedical term normalization tasks and offers an efficient alternative to more complex retrieval methods.
comment: Published by Frontiers in Digital Health
♻ ☆ Better Aligned with Survey Respondents or Training Data? Unveiling Political Leanings of LLMs on U.S. Supreme Court Cases
The increased adoption of Large Language Models (LLMs) and their potential to shape public opinion have sparked interest in assessing these models' political leanings. Building on previous research that compared LLMs and human opinions and observed political bias in system responses, we take a step further to investigate the underlying causes of such biases by empirically examining how the values and biases embedded in training corpora shape model outputs. Specifically, we propose a method to quantitatively evaluate political leanings embedded in the large pretraining corpora. Subsequently we investigate to whom are the LLMs' political leanings more aligned with, their pretrainig corpora or the surveyed human opinions. As a case study, we focus on probing the political leanings of LLMs in 32 U.S. Supreme Court cases, addressing contentious topics such as abortion and voting rights. Our findings reveal that LLMs strongly reflect the political leanings in their training data, and no strong correlation is observed with their alignment to human opinions as expressed in surveys. These results underscore the importance of responsible curation of training data and the need for robust evaluation metrics to ensure LLMs' alignment with human-centered values.
comment: under review
Computer Vision and Pattern Recognition 151
☆ ARINAR: Bi-Level Autoregressive Feature-by-Feature Generative Models
Existing autoregressive (AR) image generative models use a token-by-token generation schema. That is, they predict a per-token probability distribution and sample the next token from that distribution. The main challenge is how to model the complex distribution of high-dimensional tokens. Previous methods either are too simplistic to fit the distribution or result in slow generation speed. Instead of fitting the distribution of the whole tokens, we explore using a AR model to generate each token in a feature-by-feature way, i.e., taking the generated features as input and generating the next feature. Based on that, we propose ARINAR (AR-in-AR), a bi-level AR model. The outer AR layer take previous tokens as input, predicts a condition vector z for the next token. The inner layer, conditional on z, generates features of the next token autoregressively. In this way, the inner layer only needs to model the distribution of a single feature, for example, using a simple Gaussian Mixture Model. On the ImageNet 256x256 image generation task, ARINAR-B with 213M parameters achieves an FID of 2.75, which is comparable to the state-of-the-art MAR-B model (FID=2.31), while five times faster than the latter.
comment: Technical report. Our code is available at https://github.com/Qinyu-Allen-Zhao/Arinar
☆ SPIDER: A Comprehensive Multi-Organ Supervised Pathology Dataset and Baseline Models
Advancing AI in computational pathology requires large, high-quality, and diverse datasets, yet existing public datasets are often limited in organ diversity, class coverage, or annotation quality. To bridge this gap, we introduce SPIDER (Supervised Pathology Image-DEscription Repository), the largest publicly available patch-level dataset covering multiple organ types, including Skin, Colorectal, and Thorax, with comprehensive class coverage for each organ. SPIDER provides high-quality annotations verified by expert pathologists and includes surrounding context patches, which enhance classification performance by providing spatial context. Alongside the dataset, we present baseline models trained on SPIDER using the Hibou-L foundation model as a feature extractor combined with an attention-based classification head. The models achieve state-of-the-art performance across multiple tissue categories and serve as strong benchmarks for future digital pathology research. Beyond patch classification, the model enables rapid identification of significant areas, quantitative tissue metrics, and establishes a foundation for multimodal approaches. Both the dataset and trained models are publicly available to advance research, reproducibility, and AI-driven pathology development. Access them at: https://github.com/HistAI/SPIDER
☆ Deepfake-Eval-2024: A Multi-Modal In-the-Wild Benchmark of Deepfakes Circulated in 2024
In the age of increasingly realistic generative AI, robust deepfake detection is essential for mitigating fraud and disinformation. While many deepfake detectors report high accuracy on academic datasets, we show that these academic benchmarks are out of date and not representative of recent deepfakes. We introduce Deepfake-Eval-2024, a new deepfake detection benchmark consisting of in-the-wild deepfakes collected from social media and deepfake detection platform users in 2024. Deepfake-Eval-2024 consists of 44 hours of videos, 56.5 hours of audio, and 1,975 images, encompassing the latest manipulation technologies. The benchmark contains diverse media content from 88 different websites in 52 different languages. We find that the performance of open-source state-of-the-art deepfake detection models drops precipitously when evaluated on Deepfake-Eval-2024, with AUC decreasing by 50% for video, 48% for audio, and 45% for image models compared to previous benchmarks. We also evaluate commercial deepfake detection models and models finetuned on Deepfake-Eval-2024, and find that they have superior performance to off-the-shelf open-source models, but they do not yet reach the accuracy of human deepfake forensic analysts. The dataset is available at https://github.com/nuriachandra/Deepfake-Eval-2024.
☆ CADDI: An in-Class Activity Detection Dataset using IMU data from low-cost sensors
The monitoring and prediction of in-class student activities is of paramount importance for the comprehension of engagement and the enhancement of pedagogical efficacy. The accurate detection of these activities enables educators to modify their lessons in real time, thereby reducing negative emotional states and enhancing the overall learning experience. To this end, the use of non-intrusive devices, such as inertial measurement units (IMUs) embedded in smartwatches, represents a viable solution. The development of reliable predictive systems has been limited by the lack of large, labeled datasets in education. To bridge this gap, we present a novel dataset for in-class activity detection using affordable IMU sensors. The dataset comprises 19 diverse activities, both instantaneous and continuous, performed by 12 participants in typical classroom scenarios. It includes accelerometer, gyroscope, rotation vector data, and synchronized stereo images, offering a comprehensive resource for developing multimodal algorithms using sensor and visual data. This dataset represents a key step toward scalable solutions for activity recognition in educational settings.
☆ Multimodal Deep Learning for Subtype Classification in Breast Cancer Using Histopathological Images and Gene Expression Data
Molecular subtyping of breast cancer is crucial for personalized treatment and prognosis. Traditional classification approaches rely on either histopathological images or gene expression profiling, limiting their predictive power. In this study, we propose a deep multimodal learning framework that integrates histopathological images and gene expression data to classify breast cancer into BRCA.Luminal and BRCA.Basal / Her2 subtypes. Our approach employs a ResNet-50 model for image feature extraction and fully connected layers for gene expression processing, with a cross-attention fusion mechanism to enhance modality interaction. We conduct extensive experiments using five-fold cross-validation, demonstrating that our multimodal integration outperforms unimodal approaches in terms of classification accuracy, precision-recall AUC, and F1-score. Our findings highlight the potential of deep learning for robust and interpretable breast cancer subtype classification, paving the way for improved clinical decision-making.
comment: 9 pages, 9 figures
☆ Boltzmann Attention Sampling for Image Analysis with Small Objects
Detecting and segmenting small objects, such as lung nodules and tumor lesions, remains a critical challenge in image analysis. These objects often occupy less than 0.1% of an image, making traditional transformer architectures inefficient and prone to performance degradation due to redundant attention computations on irrelevant regions. Existing sparse attention mechanisms rely on rigid hierarchical structures, which are poorly suited for detecting small, variable, and uncertain object locations. In this paper, we propose BoltzFormer, a novel transformer-based architecture designed to address these challenges through dynamic sparse attention. BoltzFormer identifies and focuses attention on relevant areas by modeling uncertainty using a Boltzmann distribution with an annealing schedule. Initially, a higher temperature allows broader area sampling in early layers, when object location uncertainty is greatest. As the temperature decreases in later layers, attention becomes more focused, enhancing efficiency and accuracy. BoltzFormer seamlessly integrates into existing transformer architectures via a modular Boltzmann attention sampling mechanism. Comprehensive evaluations on benchmark datasets demonstrate that BoltzFormer significantly improves segmentation performance for small objects while reducing attention computation by an order of magnitude compared to previous state-of-the-art methods.
☆ In-Depth Analysis of Automated Acne Disease Recognition and Classification
Facial acne is a common disease, especially among adolescents, negatively affecting both physically and psychologically. Classifying acne is vital to providing the appropriate treatment. Traditional visual inspection or expert scanning is time-consuming and difficult to differentiate acne types. This paper introduces an automated expert system for acne recognition and classification. The proposed method employs a machine learning-based technique to classify and evaluate six types of acne diseases to facilitate the diagnosis of dermatologists. The pre-processing phase includes contrast improvement, smoothing filter, and RGB to L*a*b color conversion to eliminate noise and improve the classification accuracy. Then, a clustering-based segmentation method, k-means clustering, is applied for segmenting the disease-affected regions that pass through the feature extraction step. Characteristics of these disease-affected regions are extracted based on a combination of gray-level co-occurrence matrix (GLCM) and Statistical features. Finally, five different machine learning classifiers are employed to classify acne diseases. Experimental results show that the Random Forest (RF) achieves the highest accuracy of 98.50%, which is promising compared to the state-of-the-art methods.
☆ Developing a PET/CT Foundation Model for Cross-Modal Anatomical and Functional Imaging
In oncology, Positron Emission Tomography-Computed Tomography (PET/CT) is widely used in cancer diagnosis, staging, and treatment monitoring, as it combines anatomical details from CT with functional metabolic activity and molecular marker expression information from PET. However, existing artificial intelligence-driven PET/CT analyses rely predominantly on task-specific models trained from scratch or on limited datasets, limiting their generalizability and robustness. To address this, we propose a foundation model approach specifically designed for multimodal PET/CT imaging. We introduce the Cross-Fraternal Twin Masked Autoencoder (FratMAE), a novel framework that effectively integrates whole-body anatomical and functional or molecular information. FratMAE employs separate Vision Transformer (ViT) encoders for PET and CT scans, along with cross-attention decoders that enable synergistic interactions between modalities during masked autoencoder training. Additionally, it incorporates textual metadata to enhance PET representation learning. By pre-training on PET/CT datasets, FratMAE captures intricate cross-modal relationships and global uptake patterns, achieving superior performance on downstream tasks and demonstrating its potential as a generalizable foundation model.
comment: 11 pages, 2 figures, 3 tables
☆ MX-Font++: Mixture of Heterogeneous Aggregation Experts for Few-shot Font Generation ICASSP 2025
Few-shot Font Generation (FFG) aims to create new font libraries using limited reference glyphs, with crucial applications in digital accessibility and equity for low-resource languages, especially in multilingual artificial intelligence systems. Although existing methods have shown promising performance, transitioning to unseen characters in low-resource languages remains a significant challenge, especially when font glyphs vary considerably across training sets. MX-Font considers the content of a character from the perspective of a local component, employing a Mixture of Experts (MoE) approach to adaptively extract the component for better transition. However, the lack of a robust feature extractor prevents them from adequately decoupling content and style, leading to sub-optimal generation results. To alleviate these problems, we propose Heterogeneous Aggregation Experts (HAE), a powerful feature extraction expert that helps decouple content and style downstream from being able to aggregate information in channel and spatial dimensions. Additionally, we propose a novel content-style homogeneity loss to enhance the untangling. Extensive experiments on several datasets demonstrate that our MX-Font++ yields superior visual results in FFG and effectively outperforms state-of-the-art methods. Code and data are available at https://github.com/stephensun11/MXFontpp.
comment: 4 pages, 4 figures, accepted by ICASSP 2025
☆ A Causal Framework for Aligning Image Quality Metrics and Deep Neural Network Robustness
Image quality plays an important role in the performance of deep neural networks (DNNs) and DNNs have been widely shown to exhibit sensitivity to changes in imaging conditions. Large-scale datasets often contain images under a wide range of conditions prompting a need to quantify and understand their underlying quality distribution in order to better characterize DNN performance and robustness. Aligning the sensitivities of image quality metrics and DNNs ensures that estimates of quality can act as proxies for image/dataset difficulty independent of the task models trained/evaluated on the data. Conventional image quality assessment (IQA) seeks to measure and align quality relative to human perceptual judgments, but here we seek a quality measure that is not only sensitive to imaging conditions but also well-aligned with DNN sensitivities. We first ask whether conventional IQA metrics are also informative of DNN performance. In order to answer this question, we reframe IQA from a causal perspective and examine conditions under which quality metrics are predictive of DNN performance. We show theoretically and empirically that current IQA metrics are weak predictors of DNN performance in the context of classification. We then use our causal framework to provide an alternative formulation and a new image quality metric that is more strongly correlated with DNN performance and can act as a prior on performance without training new task models. Our approach provides a means to directly estimate the quality distribution of large-scale image datasets towards characterizing the relationship between dataset composition and DNN performance.
☆ Undertrained Image Reconstruction for Realistic Degradation in Blind Image Super-Resolution
Most super-resolution (SR) models struggle with real-world low-resolution (LR) images. This issue arises because the degradation characteristics in the synthetic datasets differ from those in real-world LR images. Since SR models are trained on pairs of high-resolution (HR) and LR images generated by downsampling, they are optimized for simple degradation. However, real-world LR images contain complex degradation caused by factors such as the imaging process and JPEG compression. Due to these differences in degradation characteristics, most SR models perform poorly on real-world LR images. This study proposes a dataset generation method using undertrained image reconstruction models. These models have the property of reconstructing low-quality images with diverse degradation from input images. By leveraging this property, this study generates LR images with diverse degradation from HR images to construct the datasets. Fine-tuning pre-trained SR models on our generated datasets improves noise removal and blur reduction, enhancing performance on real-world LR images. Furthermore, an analysis of the datasets reveals that degradation diversity contributes to performance improvements, whereas color differences between HR and LR images may degrade performance. 11 pages, (11 figures and 2 tables)
comment: 11 pages, 11 figures, 2 tables
☆ Deep Learning-Enhanced Visual Monitoring in Hazardous Underwater Environments with a Swarm of Micro-Robots ICRA 2025
Long-term monitoring and exploration of extreme environments, such as underwater storage facilities, is costly, labor-intensive, and hazardous. Automating this process with low-cost, collaborative robots can greatly improve efficiency. These robots capture images from different positions, which must be processed simultaneously to create a spatio-temporal model of the facility. In this paper, we propose a novel approach that integrates data simulation, a multi-modal deep learning network for coordinate prediction, and image reassembly to address the challenges posed by environmental disturbances causing drift and rotation in the robots' positions and orientations. Our approach enhances the precision of alignment in noisy environments by integrating visual information from snapshots, global positional context from masks, and noisy coordinates. We validate our method through extensive experiments using synthetic data that simulate real-world robotic operations in underwater settings. The results demonstrate very high coordinate prediction accuracy and plausible image assembly, indicating the real-world applicability of our approach. The assembled images provide clear and coherent views of the underwater environment for effective monitoring and inspection, showcasing the potential for broader use in extreme settings, further contributing to improved safety, efficiency, and cost reduction in hazardous field monitoring. Code is available on https://github.com/ChrisChen1023/Micro-Robot-Swarm.
comment: Accepted by ICRA 2025
☆ ArcPro: Architectural Programs for Structured 3D Abstraction of Sparse Points CVPR 2025
We introduce ArcPro, a novel learning framework built on architectural programs to recover structured 3D abstractions from highly sparse and low-quality point clouds. Specifically, we design a domain-specific language (DSL) to hierarchically represent building structures as a program, which can be efficiently converted into a mesh. We bridge feedforward and inverse procedural modeling by using a feedforward process for training data synthesis, allowing the network to make reverse predictions. We train an encoder-decoder on the points-program pairs to establish a mapping from unstructured point clouds to architectural programs, where a 3D convolutional encoder extracts point cloud features and a transformer decoder autoregressively predicts the programs in a tokenized form. Inference by our method is highly efficient and produces plausible and faithful 3D abstractions. Comprehensive experiments demonstrate that ArcPro outperforms both traditional architectural proxy reconstruction and learning-based abstraction methods. We further explore its potential to work with multi-view image and natural language inputs.
comment: CVPR 2025 (Patent Protected); Project page: https://vcc.tech/research/2025/ArcPro
☆ UAR-NVC: A Unified AutoRegressive Framework for Memory-Efficient Neural Video Compression
Implicit Neural Representations (INRs) have demonstrated significant potential in video compression by representing videos as neural networks. However, as the number of frames increases, the memory consumption for training and inference increases substantially, posing challenges in resource-constrained scenarios. Inspired by the success of traditional video compression frameworks, which process video frame by frame and can efficiently compress long videos, we adopt this modeling strategy for INRs to decrease memory consumption, while aiming to unify the frameworks from the perspective of timeline-based autoregressive modeling. In this work, we present a novel understanding of INR models from an autoregressive (AR) perspective and introduce a Unified AutoRegressive Framework for memory-efficient Neural Video Compression (UAR-NVC). UAR-NVC integrates timeline-based and INR-based neural video compression under a unified autoregressive paradigm. It partitions videos into several clips and processes each clip using a different INR model instance, leveraging the advantages of both compression frameworks while allowing seamless adaptation to either in form. To further reduce temporal redundancy between clips, we design two modules to optimize the initialization, training, and compression of these model parameters. UAR-NVC supports adjustable latencies by varying the clip length. Extensive experimental results demonstrate that UAR-NVC, with its flexible video clip setting, can adapt to resource-constrained environments and significantly improve performance compared to different baseline models.
☆ Creating Sorted Grid Layouts with Gradient-based Optimization
Visually sorted grid layouts provide an efficient method for organizing high-dimensional vectors in two-dimensional space by aligning spatial proximity with similarity relationships. This approach facilitates the effective sorting of diverse elements ranging from data points to images, and enables the simultaneous visualization of a significant number of elements. However, sorting data on two-dimensional grids is a challenge due to its high complexity. Even for a small 8-by-8 grid with 64 elements, the number of possible arrangements exceeds $1.3 \cdot 10^{89}$ - more than the number of atoms in the universe - making brute-force solutions impractical. Although various methods have been proposed to address the challenge of determining sorted grid layouts, none have investigated the potential of gradient-based optimization. In this paper, we present a novel method for grid-based sorting that exploits gradient optimization for the first time. We introduce a novel loss function that balances two opposing goals: ensuring the generation of a "valid" permutation matrix, and optimizing the arrangement on the grid to reflect the similarity between vectors, inspired by metrics that assess the quality of sorted grids. While learning-based approaches are inherently computationally complex, our method shows promising results in generating sorted grid layouts with superior sorting quality compared to existing techniques.
☆ A Joint Visual Compression and Perception Framework for Neuralmorphic Spiking Camera
The advent of neuralmorphic spike cameras has garnered significant attention for their ability to capture continuous motion with unparalleled temporal resolution.However, this imaging attribute necessitates considerable resources for binary spike data storage and transmission.In light of compression and spike-driven intelligent applications, we present the notion of Spike Coding for Intelligence (SCI), wherein spike sequences are compressed and optimized for both bit-rate and task performance.Drawing inspiration from the mammalian vision system, we propose a dual-pathway architecture for separate processing of spatial semantics and motion information, which is then merged to produce features for compression.A refinement scheme is also introduced to ensure consistency between decoded features and motion vectors.We further propose a temporal regression approach that integrates various motion dynamics, capitalizing on the advancements in warping and deformation simultaneously.Comprehensive experiments demonstrate our scheme achieves state-of-the-art (SOTA) performance for spike compression and analysis.We achieve an average 17.25% BD-rate reduction compared to SOTA codecs and a 4.3% accuracy improvement over SpiReco for spike-based classification, with 88.26% complexity reduction and 42.41% inference time saving on the encoding side.
☆ Catheter Detection and Segmentation in X-ray Images via Multi-task Learning
Automated detection and segmentation of surgical devices, such as catheters or wires, in X-ray fluoroscopic images have the potential to enhance image guidance in minimally invasive heart surgeries. In this paper, we present a convolutional neural network model that integrates a resnet architecture with multiple prediction heads to achieve real-time, accurate localization of electrodes on catheters and catheter segmentation in an end-to-end deep learning framework. We also propose a multi-task learning strategy in which our model is trained to perform both accurate electrode detection and catheter segmentation simultaneously. A key challenge with this approach is achieving optimal performance for both tasks. To address this, we introduce a novel multi-level dynamic resource prioritization method. This method dynamically adjusts sample and task weights during training to effectively prioritize more challenging tasks, where task difficulty is inversely proportional to performance and evolves throughout the training process. Experiments on both public and private datasets have demonstrated that the accuracy of our method surpasses the existing state-of-the-art methods in both single segmentation task and in the detection and segmentation multi-task. Our approach achieves a good trade-off between accuracy and efficiency, making it well-suited for real-time surgical guidance applications.
☆ Memory Efficient Continual Learning for Edge-Based Visual Anomaly Detection
Visual Anomaly Detection (VAD) is a critical task in computer vision with numerous real-world applications. However, deploying these models on edge devices presents significant challenges, such as constrained computational and memory resources. Additionally, dynamic data distributions in real-world settings necessitate continuous model adaptation, further complicating deployment under limited resources. To address these challenges, we present a novel investigation into the problem of Continual Learning for Visual Anomaly Detection (CLAD) on edge devices. We evaluate the STFPM approach, given its low memory footprint on edge devices, which demonstrates good performance when combined with the Replay approach. Furthermore, we propose to study the behavior of a recently proposed approach, PaSTe, specifically designed for the edge but not yet explored in the Continual Learning context. Our results show that PaSTe is not only a lighter version of STPFM, but it also achieves superior anomaly detection performance, improving the f1 pixel performance by 10% with the Replay technique. In particular, the structure of PaSTe allows us to test it using a series of Compressed Replay techniques, reducing memory overhead by a maximum of 91.5% compared to the traditional Replay for STFPM. Our study proves the feasibility of deploying VAD models that adapt and learn incrementally on CLAD scenarios on resource-constrained edge devices.
☆ STAA-SNN: Spatial-Temporal Attention Aggregator for Spiking Neural Networks CVPR 2025
Spiking Neural Networks (SNNs) have gained significant attention due to their biological plausibility and energy efficiency, making them promising alternatives to Artificial Neural Networks (ANNs). However, the performance gap between SNNs and ANNs remains a substantial challenge hindering the widespread adoption of SNNs. In this paper, we propose a Spatial-Temporal Attention Aggregator SNN (STAA-SNN) framework, which dynamically focuses on and captures both spatial and temporal dependencies. First, we introduce a spike-driven self-attention mechanism specifically designed for SNNs. Additionally, we pioneeringly incorporate position encoding to integrate latent temporal relationships into the incoming features. For spatial-temporal information aggregation, we employ step attention to selectively amplify relevant features at different steps. Finally, we implement a time-step random dropout strategy to avoid local optima. As a result, STAA-SNN effectively captures both spatial and temporal dependencies, enabling the model to analyze complex patterns and make accurate predictions. The framework demonstrates exceptional performance across diverse datasets and exhibits strong generalization capabilities. Notably, STAA-SNN achieves state-of-the-art results on neuromorphic datasets CIFAR10-DVS, with remarkable performances of 97.14%, 82.05% and 70.40% on the static datasets CIFAR-10, CIFAR-100 and ImageNet, respectively. Furthermore, our model exhibits improved performance ranging from 0.33\% to 2.80\% with fewer time steps. The code for the model is available on GitHub.
comment: Accepted by CVPR 2025
☆ Class-Aware PillarMix: Can Mixed Sample Data Augmentation Enhance 3D Object Detection with Radar Point Clouds? IROS 2025
Due to the significant effort required for data collection and annotation in 3D perception tasks, mixed sample data augmentation (MSDA) has been widely studied to generate diverse training samples by mixing existing data. Recently, many MSDA techniques have been developed for point clouds, but they mainly target LiDAR data, leaving their application to radar point clouds largely unexplored. In this paper, we examine the feasibility of applying existing MSDA methods to radar point clouds and identify several challenges in adapting these techniques. These obstacles stem from the radar's irregular angular distribution, deviations from a single-sensor polar layout in multi-radar setups, and point sparsity. To address these issues, we propose Class-Aware PillarMix (CAPMix), a novel MSDA approach that applies MixUp at the pillar level in 3D point clouds, guided by class labels. Unlike methods that rely a single mix ratio to the entire sample, CAPMix assigns an independent ratio to each pillar, boosting sample diversity. To account for the density of different classes, we use class-specific distributions: for dense objects (e.g., large vehicles), we skew ratios to favor points from another sample, while for sparse objects (e.g., pedestrians), we sample more points from the original. This class-aware mixing retains critical details and enriches each sample with new information, ultimately generating more diverse training data. Experimental results demonstrate that our method not only significantly boosts performance but also outperforms existing MSDA approaches across two datasets (Bosch Street and K-Radar). We believe that this straightforward yet effective approach will spark further investigation into MSDA techniques for radar data.
comment: 8 pages, 6 figures, 4 tables, submitted to 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025)
☆ TReND: Transformer derived features and Regularized NMF for neonatal functional network Delineation
Precise parcellation of functional networks (FNs) of early developing human brain is the fundamental basis for identifying biomarker of developmental disorders and understanding functional development. Resting-state fMRI (rs-fMRI) enables in vivo exploration of functional changes, but adult FN parcellations cannot be directly applied to the neonates due to incomplete network maturation. No standardized neonatal functional atlas is currently available. To solve this fundamental issue, we propose TReND, a novel and fully automated self-supervised transformer-autoencoder framework that integrates regularized nonnegative matrix factorization (RNMF) to unveil the FNs in neonates. TReND effectively disentangles spatiotemporal features in voxel-wise rs-fMRI data. The framework integrates confidence-adaptive masks into transformer self-attention layers to mitigate noise influence. A self supervised decoder acts as a regulator to refine the encoder's latent embeddings, which serve as reliable temporal features. For spatial coherence, we incorporate brain surface-based geodesic distances as spatial encodings along with functional connectivity from temporal features. The TReND clustering approach processes these features under sparsity and smoothness constraints, producing robust and biologically plausible parcellations. We extensively validated our TReND framework on three different rs-fMRI datasets: simulated, dHCP and HCP-YA against comparable traditional feature extraction and clustering techniques. Our results demonstrated the superiority of the TReND framework in the delineation of neonate FNs with significantly better spatial contiguity and functional homogeneity. Collectively, we established TReND, a novel and robust framework, for neonatal FN delineation. TReND-derived neonatal FNs could serve as a neonatal functional atlas for perinatal populations in health and disease.
comment: 10 Pages, 5 figures
☆ State of play and future directions in industrial computer vision AI standards
The recent tremendous advancements in the areas of Artificial Intelligence (AI) and Deep Learning (DL) have also resulted into corresponding remarkable progress in the field of Computer Vision (CV), showcasing robust technological solutions in a wide range of application sectors of high industrial interest (e.g., healthcare, autonomous driving, automation, etc.). Despite the outstanding performance of CV systems in specific domains, their development and exploitation at industrial-scale necessitates, among other, the addressing of requirements related to the reliability, transparency, trustworthiness, security, safety, and robustness of the developed AI models. The latter raises the imperative need for the development of efficient, comprehensive and widely-adopted industrial standards. In this context, this study investigates the current state of play regarding the development of industrial computer vision AI standards, emphasizing on critical aspects, like model interpretability, data quality, and regulatory compliance. In particular, a systematic analysis of launched and currently developing CV standards, proposed by the main international standardization bodies (e.g. ISO/IEC, IEEE, DIN, etc.) is performed. The latter is complemented by a comprehensive discussion on the current challenges and future directions observed in this regularization endeavor.
☆ 10K is Enough: An Ultra-Lightweight Binarized Network for Infrared Small-Target Detection
The widespread deployment of InfRared Small-Target Detection(IRSTD) algorithms on edge devices necessitates the exploration of model compression techniques. Binary neural networks (BNNs) are distinguished by their exceptional efficiency in model compression. However, the small size of infrared targets introduces stringent precision requirements for the IRSTD task, while the inherent precision loss during binarization presents a significant challenge. To address this, we propose the Binarized Infrared Small-Target Detection Network (BiisNet), which preserves the core operations of binarized convolutions while integrating full-precision features into the network's information flow. Specifically, we propose the Dot-Binary Convolution, which retains fine-grained semantic information in feature maps while still leveraging the binarized convolution operations. In addition, we introduce a smooth and adaptive Dynamic Softsign function, which provides more comprehensive and progressively finer gradient during back-propagation, enhancing model stability and promoting an optimal weight distribution.Experimental results demonstrate that BiisNet not only significantly outperforms other binary architectures but also demonstrates strong competitiveness among state-of-the-art full-precision models.
☆ A dataset-free approach for self-supervised learning of 3D reflectional symmetries
In this paper, we explore a self-supervised model that learns to detect the symmetry of a single object without requiring a dataset-relying solely on the input object itself. We hypothesize that the symmetry of an object can be determined by its intrinsic features, eliminating the need for large datasets during training. Additionally, we design a self-supervised learning strategy that removes the necessity of ground truth labels. These two key elements make our approach both effective and efficient, addressing the prohibitive costs associated with constructing large, labeled datasets for this task. The novelty of our method lies in computing features for each point on the object based on the idea that symmetric points should exhibit similar visual appearances. To achieve this, we leverage features extracted from a foundational image model to compute a visual descriptor for the points. This approach equips the point cloud with visual features that facilitate the optimization of our self-supervised model. Experimental results demonstrate that our method surpasses the state-of-the-art models trained on large datasets. Furthermore, our model is more efficient, effective, and operates with minimal computational and data resources.
☆ XFMamba: Cross-Fusion Mamba for Multi-View Medical Image Classification
Compared to single view medical image classification, using multiple views can significantly enhance predictive accuracy as it can account for the complementarity of each view while leveraging correlations between views. Existing multi-view approaches typically employ separate convolutional or transformer branches combined with simplistic feature fusion strategies. However, these approaches inadvertently disregard essential cross-view correlations, leading to suboptimal classification performance, and suffer from challenges with limited receptive field (CNNs) or quadratic computational complexity (transformers). Inspired by state space sequence models, we propose XFMamba, a pure Mamba-based cross-fusion architecture to address the challenge of multi-view medical image classification. XFMamba introduces a novel two-stage fusion strategy, facilitating the learning of single-view features and their cross-view disparity. This mechanism captures spatially long-range dependencies in each view while enhancing seamless information transfer between views. Results on three public datasets, MURA, CheXpert and DDSM, illustrate the effectiveness of our approach across diverse multi-view medical image classification tasks, showing that it outperforms existing convolution-based and transformer-based multi-view methods. Code is available at https://github.com/XZheng0427/XFMamba.
☆ ZAPBench: A Benchmark for Whole-Brain Activity Prediction in Zebrafish
Data-driven benchmarks have led to significant progress in key scientific modeling domains including weather and structural biology. Here, we introduce the Zebrafish Activity Prediction Benchmark (ZAPBench) to measure progress on the problem of predicting cellular-resolution neural activity throughout an entire vertebrate brain. The benchmark is based on a novel dataset containing 4d light-sheet microscopy recordings of over 70,000 neurons in a larval zebrafish brain, along with motion stabilized and voxel-level cell segmentations of these data that facilitate development of a variety of forecasting methods. Initial results from a selection of time series and volumetric video modeling approaches achieve better performance than naive baseline methods, but also show room for further improvement. The specific brain used in the activity recording is also undergoing synaptic-level anatomical mapping, which will enable future integration of detailed structural information into forecasting methods.
☆ Smoothing the Shift: Towards Stable Test-Time Adaptation under Complex Multimodal Noises ICLR 2025
Test-Time Adaptation (TTA) aims to tackle distribution shifts using unlabeled test data without access to the source data. In the context of multimodal data, there are more complex noise patterns than unimodal data such as simultaneous corruptions for multiple modalities and missing modalities. Besides, in real-world applications, corruptions from different distribution shifts are always mixed. Existing TTA methods always fail in such multimodal scenario because the abrupt distribution shifts will destroy the prior knowledge from the source model, thus leading to performance degradation. To this end, we reveal a new challenge named multimodal wild TTA. To address this challenging problem, we propose two novel strategies: sample identification with interquartile range Smoothing and unimodal assistance, and Mutual information sharing (SuMi). SuMi smooths the adaptation process by interquartile range which avoids the abrupt distribution shifts. Then, SuMi fully utilizes the unimodal features to select low-entropy samples with rich multimodal information for optimization. Furthermore, mutual information sharing is introduced to align the information, reduce the discrepancies and enhance the information utilization across different modalities. Extensive experiments on two public datasets show the effectiveness and superiority over existing methods under the complex noise patterns in multimodal data. Code is available at https://github.com/zrguo/SuMi.
comment: Accepted at ICLR 2025
☆ ARC-Flow : Articulated, Resolution-Agnostic, Correspondence-Free Matching and Interpolation of 3D Shapes Under Flow Fields
This work presents a unified framework for the unsupervised prediction of physically plausible interpolations between two 3D articulated shapes and the automatic estimation of dense correspondence between them. Interpolation is modelled as a diffeomorphic transformation using a smooth, time-varying flow field governed by Neural Ordinary Differential Equations (ODEs). This ensures topological consistency and non-intersecting trajectories while accommodating hard constraints, such as volume preservation, and soft constraints, \eg physical priors. Correspondence is recovered using an efficient Varifold formulation, that is effective on high-fidelity surfaces with differing parameterisations. By providing a simple skeleton for the source shape only, we impose physically motivated constraints on the deformation field and resolve symmetric ambiguities. This is achieved without relying on skinning weights or any prior knowledge of the skeleton's target pose configuration. Qualitative and quantitative results demonstrate competitive or superior performance over existing state-of-the-art approaches in both shape correspondence and interpolation tasks across standard datasets.
comment: 11 pages, 6 figures
☆ Resource-Efficient Affordance Grounding with Complementary Depth and Semantic Prompts
Affordance refers to the functional properties that an agent perceives and utilizes from its environment, and is key perceptual information required for robots to perform actions. This information is rich and multimodal in nature. Existing multimodal affordance methods face limitations in extracting useful information, mainly due to simple structural designs, basic fusion methods, and large model parameters, making it difficult to meet the performance requirements for practical deployment. To address these issues, this paper proposes the BiT-Align image-depth-text affordance mapping framework. The framework includes a Bypass Prompt Module (BPM) and a Text Feature Guidance (TFG) attention selection mechanism. BPM integrates the auxiliary modality depth image directly as a prompt to the primary modality RGB image, embedding it into the primary modality encoder without introducing additional encoders. This reduces the model's parameter count and effectively improves functional region localization accuracy. The TFG mechanism guides the selection and enhancement of attention heads in the image encoder using textual features, improving the understanding of affordance characteristics. Experimental results demonstrate that the proposed method achieves significant performance improvements on public AGD20K and HICO-IIF datasets. On the AGD20K dataset, compared with the current state-of-the-art method, we achieve a 6.0% improvement in the KLD metric, while reducing model parameters by 88.8%, demonstrating practical application values. The source code will be made publicly available at https://github.com/DAWDSE/BiT-Align.
comment: The source code will be made publicly available at https://github.com/DAWDSE/BiT-Align
☆ Seeing is Understanding: Unlocking Causal Attention into Modality-Mutual Attention for Multimodal LLMs
Recent Multimodal Large Language Models (MLLMs) have demonstrated significant progress in perceiving and reasoning over multimodal inquiries, ushering in a new research era for foundation models. However, vision-language misalignment in MLLMs has emerged as a critical challenge, where the textual responses generated by these models are not factually aligned with the given text-image inputs. Existing efforts to address vision-language misalignment have focused on developing specialized vision-language connectors or leveraging visual instruction tuning from diverse domains. In this paper, we tackle this issue from a fundamental yet unexplored perspective by revisiting the core architecture of MLLMs. Most MLLMs are typically built on decoder-only LLMs consisting of a causal attention mechanism, which limits the ability of earlier modalities (e.g., images) to incorporate information from later modalities (e.g., text). To address this problem, we propose AKI, a novel MLLM that unlocks causal attention into modality-mutual attention (MMA) to enable image tokens to attend to text tokens. This simple yet effective design allows AKI to achieve superior performance in 12 multimodal understanding benchmarks (+7.2% on average) without introducing additional parameters and increasing training time. Our MMA design is intended to be generic, allowing for application across various modalities, and scalable to accommodate diverse multimodal scenarios. The code is publicly available at https://github.com/sony/aki, and we will release our AKI-4B model to encourage further advancements in MLLMs across various directions.
comment: Preprint
☆ StageDesigner: Artistic Stage Generation for Scenography via Theater Scripts
In this work, we introduce StageDesigner, the first comprehensive framework for artistic stage generation using large language models combined with layout-controlled diffusion models. Given the professional requirements of stage scenography, StageDesigner simulates the workflows of seasoned artists to generate immersive 3D stage scenes. Specifically, our approach is divided into three primary modules: Script Analysis, which extracts thematic and spatial cues from input scripts; Foreground Generation, which constructs and arranges essential 3D objects; and Background Generation, which produces a harmonious background aligned with the narrative atmosphere and maintains spatial coherence by managing occlusions between foreground and background elements. Furthermore, we introduce the StagePro-V1 dataset, a dedicated dataset with 276 unique stage scenes spanning different historical styles and annotated with scripts, images, and detailed 3D layouts, specifically tailored for this task. Finally, evaluations using both standard and newly proposed metrics, along with extensive user studies, demonstrate the effectiveness of StageDesigner. Project can be found at: https://deadsmither5.github.io/2025/01/03/StageDesigner/
☆ CMMLoc: Advancing Text-to-PointCloud Localization with Cauchy-Mixture-Model Based Framework
The goal of point cloud localization based on linguistic description is to identify a 3D position using textual description in large urban environments, which has potential applications in various fields, such as determining the location for vehicle pickup or goods delivery. Ideally, for a textual description and its corresponding 3D location, the objects around the 3D location should be fully described in the text description. However, in practical scenarios, e.g., vehicle pickup, passengers usually describe only the part of the most significant and nearby surroundings instead of the entire environment. In response to this $\textbf{partially relevant}$ challenge, we propose $\textbf{CMMLoc}$, an uncertainty-aware $\textbf{C}$auchy-$\textbf{M}$ixture-$\textbf{M}$odel ($\textbf{CMM}$) based framework for text-to-point-cloud $\textbf{Loc}$alization. To model the uncertain semantic relations between text and point cloud, we integrate CMM constraints as a prior during the interaction between the two modalities. We further design a spatial consolidation scheme to enable adaptive aggregation of different 3D objects with varying receptive fields. To achieve precise localization, we propose a cardinal direction integration module alongside a modality pre-alignment strategy, helping capture the spatial relationships among objects and bringing the 3D objects closer to the text modality. Comprehensive experiments validate that CMMLoc outperforms existing methods, achieving state-of-the-art results on the KITTI360Pose dataset. Codes are available in this GitHub repository https://github.com/kevin301342/CMMLoc.
☆ A Hypernetwork-Based Approach to KAN Representation of Audio Signals
Implicit neural representations (INR) have gained prominence for efficiently encoding multimedia data, yet their applications in audio signals remain limited. This study introduces the Kolmogorov-Arnold Network (KAN), a novel architecture using learnable activation functions, as an effective INR model for audio representation. KAN demonstrates superior perceptual performance over previous INRs, achieving the lowest Log-SpectralDistance of 1.29 and the highest Perceptual Evaluation of Speech Quality of 3.57 for 1.5 s audio. To extend KAN's utility, we propose FewSound, a hypernetwork-based architecture that enhances INR parameter updates. FewSound outperforms the state-of-the-art HyperSound, with a 33.3% improvement in MSE and 60.87% in SI-SNR. These results show KAN as a robust and adaptable audio representation with the potential for scalability and integration into various hypernetwork frameworks. The source code can be accessed at https://github.com/gmum/fewsound.git.
☆ Unveiling the Potential of Segment Anything Model 2 for RGB-Thermal Semantic Segmentation with Language Guidance
The perception capability of robotic systems relies on the richness of the dataset. Although Segment Anything Model 2 (SAM2), trained on large datasets, demonstrates strong perception potential in perception tasks, its inherent training paradigm prevents it from being suitable for RGB-T tasks. To address these challenges, we propose SHIFNet, a novel SAM2-driven Hybrid Interaction Paradigm that unlocks the potential of SAM2 with linguistic guidance for efficient RGB-Thermal perception. Our framework consists of two key components: (1) Semantic-Aware Cross-modal Fusion (SACF) module that dynamically balances modality contributions through text-guided affinity learning, overcoming SAM2's inherent RGB bias; (2) Heterogeneous Prompting Decoder (HPD) that enhances global semantic information through a semantic enhancement module and then combined with category embeddings to amplify cross-modal semantic consistency. With 32.27M trainable parameters, SHIFNet achieves state-of-the-art segmentation performance on public benchmarks, reaching 89.8% on PST900 and 67.8% on FMB, respectively. The framework facilitates the adaptation of pre-trained large models to RGB-T segmentation tasks, effectively mitigating the high costs associated with data collection while endowing robotic systems with comprehensive perception capabilities. The source code will be made publicly available at https://github.com/iAsakiT3T/SHIFNet.
comment: The source code will be made publicly available at https://github.com/iAsakiT3T/SHIFNet
☆ MM-OR: A Large Multimodal Operating Room Dataset for Semantic Understanding of High-Intensity Surgical Environments
Operating rooms (ORs) are complex, high-stakes environments requiring precise understanding of interactions among medical staff, tools, and equipment for enhancing surgical assistance, situational awareness, and patient safety. Current datasets fall short in scale, realism and do not capture the multimodal nature of OR scenes, limiting progress in OR modeling. To this end, we introduce MM-OR, a realistic and large-scale multimodal spatiotemporal OR dataset, and the first dataset to enable multimodal scene graph generation. MM-OR captures comprehensive OR scenes containing RGB-D data, detail views, audio, speech transcripts, robotic logs, and tracking data and is annotated with panoptic segmentations, semantic scene graphs, and downstream task labels. Further, we propose MM2SG, the first multimodal large vision-language model for scene graph generation, and through extensive experiments, demonstrate its ability to effectively leverage multimodal inputs. Together, MM-OR and MM2SG establish a new benchmark for holistic OR understanding, and open the path towards multimodal scene analysis in complex, high-stakes environments. Our code, and data is available at https://github.com/egeozsoy/MM-OR.
☆ TS-CGNet: Temporal-Spatial Fusion Meets Centerline-Guided Diffusion for BEV Mapping
Bird's Eye View (BEV) perception technology is crucial for autonomous driving, as it generates top-down 2D maps for environment perception, navigation, and decision-making. Nevertheless, the majority of current BEV map generation studies focusing on visual map generation lack depth-aware reasoning capabilities. They exhibit limited efficacy in managing occlusions and handling complex environments, with a notable decline in perceptual performance under adverse weather conditions or low-light scenarios. Therefore, this paper proposes TS-CGNet, which leverages Temporal-Spatial fusion with Centerline-Guided diffusion. This visual framework, grounded in prior knowledge, is designed for integration into any existing network for building BEV maps. Specifically, this framework is decoupled into three parts: Local mapping system involves the initial generation of semantic maps using purely visual information; The Temporal-Spatial Aligner Module (TSAM) integrates historical information into mapping generation by applying transformation matrices; The Centerline-Guided Diffusion Model (CGDM) is a prediction module based on the diffusion model. CGDM incorporates centerline information through spatial-attention mechanisms to enhance semantic segmentation reconstruction. We construct BEV semantic segmentation maps by our methods on the public nuScenes and the robustness benchmarks under various corruptions. Our method improves 1.90%, 1.73%, and 2.87% for perceived ranges of 60x30m, 120x60m, and 240x60m in the task of BEV HD mapping. TS-CGNet attains an improvement of 1.92% for perceived ranges of 100x100m in the task of BEV semantic mapping. Moreover, TS-CGNet achieves an average improvement of 2.92% in detection accuracy under varying weather conditions and sensor interferences in the perception range of 240x60m. The source code will be publicly available at https://github.com/krabs-H/TS-CGNet.
comment: The source code will be publicly available at https://github.com/krabs-H/TS-CGNet
☆ SPG: Improving Motion Diffusion by Smooth Perturbation Guidance
This paper presents a test-time guidance method to improve the output quality of the human motion diffusion models without requiring additional training. To have negative guidance, Smooth Perturbation Guidance (SPG) builds a weak model by temporally smoothing the motion in the denoising steps. Compared to model-agnostic methods originating from the image generation field, SPG effectively mitigates out-of-distribution issues when perturbing motion diffusion models. In SPG guidance, the nature of motion structure remains intact. This work conducts a comprehensive analysis across distinct model architectures and tasks. Despite its extremely simple implementation and no need for additional training requirements, SPG consistently enhances motion fidelity. Project page can be found at https://spg-blind.vercel.app/
☆ Tracking-Aware Deformation Field Estimation for Non-rigid 3D Reconstruction in Robotic Surgeries
Minimally invasive procedures have been advanced rapidly by the robotic laparoscopic surgery. The latter greatly assists surgeons in sophisticated and precise operations with reduced invasiveness. Nevertheless, it is still safety critical to be aware of even the least tissue deformation during instrument-tissue interactions, especially in 3D space. To address this, recent works rely on NeRF to render 2D videos from different perspectives and eliminate occlusions. However, most of the methods fail to predict the accurate 3D shapes and associated deformation estimates robustly. Differently, we propose Tracking-Aware Deformation Field (TADF), a novel framework which reconstructs the 3D mesh along with the 3D tissue deformation simultaneously. It first tracks the key points of soft tissue by a foundation vision model, providing an accurate 2D deformation field. Then, the 2D deformation field is smoothly incorporated with a neural implicit reconstruction network to obtain tissue deformation in the 3D space. Finally, we experimentally demonstrate that the proposed method provides more accurate deformation estimation compared with other 3D neural reconstruction methods in two public datasets.
☆ Towards a robust R2D2 paradigm for radio-interferometric imaging: revisiting DNN training and architecture
The R2D2 Deep Neural Network (DNN) series was recently introduced for image formation in radio interferometry. It can be understood as a learned version of CLEAN, whose minor cycles are substituted with DNNs. We revisit R2D2 on the grounds of series convergence, training methodology, and DNN architecture, improving its robustness in terms of generalisability beyond training conditions, capability to deliver high data fidelity, and epistemic uncertainty. Firstly, while still focusing on telescope-specific training, we enhance the learning process by randomising Fourier sampling integration times, incorporating multi-scan multi-noise configurations, and varying imaging settings, including pixel resolution and visibility-weighting scheme. Secondly, we introduce a convergence criterion whereby the reconstruction process stops when the data residual is compatible with noise, rather than simply using all available DNNs. This not only increases the reconstruction efficiency by reducing its computational cost, but also refines training by pruning out the data/image pairs for which optimal data fidelity is reached before training the next DNN. Thirdly, we substitute R2D2's early U-Net DNN with a novel architecture (U-WDSR) combining U-Net and WDSR, which leverages wide activation, dense connections, weight normalisation, and low-rank convolution to improve feature reuse and reconstruction precision. As previously, R2D2 was trained for monochromatic intensity imaging with the Very Large Array (VLA) at fixed $512 \times 512$ image size. Simulations on a wide range of inverse problems and a case study on real data reveal that the new R2D2 model consistently outperforms its earlier version in image reconstruction quality, data fidelity, and epistemic uncertainty.
comment: 17 pages, 6 figures
☆ Federated nnU-Net for Privacy-Preserving Medical Image Segmentation
The nnU-Net framework has played a crucial role in medical image segmentation and has become the gold standard in multitudes of applications targeting different diseases, organs, and modalities. However, so far it has been used primarily in a centralized approach where the data collected from hospitals are stored in one center and used to train the nnU-Net. This centralized approach has various limitations, such as leakage of sensitive patient information and violation of patient privacy. Federated learning is one of the approaches to train a segmentation model in a decentralized manner that helps preserve patient privacy. In this paper, we propose FednnU-Net, a federated learning extension of nnU-Net. We introduce two novel federated learning methods to the nnU-Net framework - Federated Fingerprint Extraction (FFE) and Asymmetric Federated Averaging (AsymFedAvg) - and experimentally show their consistent performance for breast, cardiac and fetal segmentation using 6 datasets representing samples from 18 institutions. Additionally, to further promote research and deployment of decentralized training in privacy constrained institutions, we make our plug-n-play framework public. The source-code is available at https://github.com/faildeny/FednnUNet .
comment: In review
☆ PVTree: Realistic and Controllable Palm Vein Generation for Recognition Tasks
Palm vein recognition is an emerging biometric technology that offers enhanced security and privacy. However, acquiring sufficient palm vein data for training deep learning-based recognition models is challenging due to the high costs of data collection and privacy protection constraints. This has led to a growing interest in generating pseudo-palm vein data using generative models. Existing methods, however, often produce unrealistic palm vein patterns or struggle with controlling identity and style attributes. To address these issues, we propose a novel palm vein generation framework named PVTree. First, the palm vein identity is defined by a complex and authentic 3D palm vascular tree, created using an improved Constrained Constructive Optimization (CCO) algorithm. Second, palm vein patterns of the same identity are generated by projecting the same 3D vascular tree into 2D images from different views and converting them into realistic images using a generative model. As a result, PVTree satisfies the need for both identity consistency and intra-class diversity. Extensive experiments conducted on several publicly available datasets demonstrate that our proposed palm vein generation method surpasses existing methods and achieves a higher TAR@FAR=1e-4 under the 1:1 Open-set protocol. To the best of our knowledge, this is the first time that the performance of a recognition model trained on synthetic palm vein data exceeds that of the recognition model trained on real data, which indicates that palm vein image generation research has a promising future.
☆ RectifiedHR: Enable Efficient High-Resolution Image Generation via Energy Rectification
Diffusion models have achieved remarkable advances in various image generation tasks. However, their performance notably declines when generating images at resolutions higher than those used during the training period. Despite the existence of numerous methods for producing high-resolution images, they either suffer from inefficiency or are hindered by complex operations. In this paper, we propose RectifiedHR, an efficient and straightforward solution for training-free high-resolution image generation. Specifically, we introduce the noise refresh strategy, which theoretically only requires a few lines of code to unlock the model's high-resolution generation ability and improve efficiency. Additionally, we first observe the phenomenon of energy decay that may cause image blurriness during the high-resolution image generation process. To address this issue, we propose an Energy Rectification strategy, where modifying the hyperparameters of the classifier-free guidance effectively improves the generation performance. Our method is entirely training-free and boasts a simple implementation logic. Through extensive comparisons with numerous baseline methods, our RectifiedHR demonstrates superior effectiveness and efficiency.
comment: Project Page: https://zhenyangcs.github.io/RectifiedHR-Diffusion/
☆ TeTRA-VPR: A Ternary Transformer Approach for Compact Visual Place Recognition
Visual Place Recognition (VPR) localizes a query image by matching it against a database of geo-tagged reference images, making it essential for navigation and mapping in robotics. Although Vision Transformer (ViT) solutions deliver high accuracy, their large models often exceed the memory and compute budgets of resource-constrained platforms such as drones and mobile robots. To address this issue, we propose TeTRA, a ternary transformer approach that progressively quantizes the ViT backbone to 2-bit precision and binarizes its final embedding layer, offering substantial reductions in model size and latency. A carefully designed progressive distillation strategy preserves the representational power of a full-precision teacher, allowing TeTRA to retain or even surpass the accuracy of uncompressed convolutional counterparts, despite using fewer resources. Experiments on standard VPR benchmarks demonstrate that TeTRA reduces memory consumption by up to 69% compared to efficient baselines, while lowering inference latency by 35%, with either no loss or a slight improvement in recall@1. These gains enable high-accuracy VPR on power-constrained, memory-limited robotic platforms, making TeTRA an appealing solution for real-world deployment.
☆ Remote Sensing Image Classification Using Convolutional Neural Network (CNN) and Transfer Learning Techniques
This study investigates the classification of aerial images depicting transmission towers, forests, farmland, and mountains. To complete the classification job, features are extracted from input photos using a Convolutional Neural Network (CNN) architecture. Then, the images are classified using Softmax. To test the model, we ran it for ten epochs using a batch size of 90, the Adam optimizer, and a learning rate of 0.001. Both training and assessment are conducted using a dataset that blends self-collected pictures from Google satellite imagery with the MLRNet dataset. The comprehensive dataset comprises 10,400 images. Our study shows that transfer learning models and MobileNetV2 in particular, work well for landscape categorization. These models are good options for practical use because they strike a good mix between precision and efficiency; our approach achieves results with an overall accuracy of 87% on the built CNN model. Furthermore, we reach even higher accuracies by utilizing the pretrained VGG16 and MobileNetV2 models as a starting point for transfer learning. Specifically, VGG16 achieves an accuracy of 90% and a test loss of 0.298, while MobileNetV2 outperforms both models with an accuracy of 96% and a test loss of 0.119; the results demonstrate the effectiveness of employing transfer learning with MobileNetV2 for classifying transmission towers, forests, farmland, and mountains.
comment: This paper is published in Journal of Computer Science, Volume 21 No. 3, 2025. It contains 635-645 pages
☆ Q&C: When Quantization Meets Cache in Efficient Image Generation
Quantization and cache mechanisms are typically applied individually for efficient Diffusion Transformers (DiTs), each demonstrating notable potential for acceleration. However, the promoting effect of combining the two mechanisms on efficient generation remains under-explored. Through empirical investigation, we find that the combination of quantization and cache mechanisms for DiT is not straightforward, and two key challenges lead to severe catastrophic performance degradation: (i) the sample efficacy of calibration datasets in post-training quantization (PTQ) is significantly eliminated by cache operation; (ii) the combination of the above mechanisms introduces more severe exposure bias within sampling distribution, resulting in amplified error accumulation in the image generation process. In this work, we take advantage of these two acceleration mechanisms and propose a hybrid acceleration method by tackling the above challenges, aiming to further improve the efficiency of DiTs while maintaining excellent generation capability. Concretely, a temporal-aware parallel clustering (TAP) is designed to dynamically improve the sample selection efficacy for the calibration within PTQ for different diffusion steps. A variance compensation (VC) strategy is derived to correct the sampling distribution. It mitigates exposure bias through an adaptive correction factor generation. Extensive experiments have shown that our method has accelerated DiTs by 12.7x while preserving competitive generation capability. The code will be available at https://github.com/xinding-sys/Quant-Cache.
comment: 11 pages
☆ ROCKET-2: Steering Visuomotor Policy via Cross-View Goal Alignment
We aim to develop a goal specification method that is semantically clear, spatially sensitive, and intuitive for human users to guide agent interactions in embodied environments. Specifically, we propose a novel cross-view goal alignment framework that allows users to specify target objects using segmentation masks from their own camera views rather than the agent's observations. We highlight that behavior cloning alone fails to align the agent's behavior with human intent when the human and agent camera views differ significantly. To address this, we introduce two auxiliary objectives: cross-view consistency loss and target visibility loss, which explicitly enhance the agent's spatial reasoning ability. According to this, we develop ROCKET-2, a state-of-the-art agent trained in Minecraft, achieving an improvement in the efficiency of inference 3x to 6x. We show ROCKET-2 can directly interpret goals from human camera views for the first time, paving the way for better human-agent interaction.
☆ Deepfake Detection via Knowledge Injection
Deepfake detection technologies become vital because current generative AI models can generate realistic deepfakes, which may be utilized in malicious purposes. Existing deepfake detection methods either rely on developing classification methods to better fit the distributions of the training data, or exploiting forgery synthesis mechanisms to learn a more comprehensive forgery distribution. Unfortunately, these methods tend to overlook the essential role of real data knowledge, which limits their generalization ability in processing the unseen real and fake data. To tackle these challenges, in this paper, we propose a simple and novel approach, named Knowledge Injection based deepfake Detection (KID), by constructing a multi-task learning based knowledge injection framework, which can be easily plugged into existing ViT-based backbone models, including foundation models. Specifically, a knowledge injection module is proposed to learn and inject necessary knowledge into the backbone model, to achieve a more accurate modeling of the distributions of real and fake data. A coarse-grained forgery localization branch is constructed to learn the forgery locations in a multi-task learning manner, to enrich the learned forgery knowledge for the knowledge injection module. Two layer-wise suppression and contrast losses are proposed to emphasize the knowledge of real data in the knowledge injection module, to further balance the portions of the real and fake knowledge. Extensive experiments have demonstrated that our KID possesses excellent compatibility with different scales of Vit-based backbone models, and achieves state-of-the-art generalization performance while enhancing the training convergence speed.
☆ Joint Out-of-Distribution Filtering and Data Discovery Active Learning
As the data demand for deep learning models increases, active learning (AL) becomes essential to strategically select samples for labeling, which maximizes data efficiency and reduces training costs. Real-world scenarios necessitate the consideration of incomplete data knowledge within AL. Prior works address handling out-of-distribution (OOD) data, while another research direction has focused on category discovery. However, a combined analysis of real-world considerations combining AL with out-of-distribution data and category discovery remains unexplored. To address this gap, we propose Joint Out-of-distribution filtering and data Discovery Active learning (Joda) , to uniquely address both challenges simultaneously by filtering out OOD data before selecting candidates for labeling. In contrast to previous methods, we deeply entangle the training procedure with filter and selection to construct a common feature space that aligns known and novel categories while separating OOD samples. Unlike previous works, Joda is highly efficient and completely omits auxiliary models and training access to the unlabeled pool for filtering or selection. In extensive experiments on 18 configurations and 3 metrics, \ours{} consistently achieves the highest accuracy with the best class discovery to OOD filtering balance compared to state-of-the-art competitor approaches.
☆ Deep Robust Reversible Watermarking
Robust Reversible Watermarking (RRW) enables perfect recovery of cover images and watermarks in lossless channels while ensuring robust watermark extraction in lossy channels. Existing RRW methods, mostly non-deep learning-based, face complex designs, high computational costs, and poor robustness, limiting their practical use. This paper proposes Deep Robust Reversible Watermarking (DRRW), a deep learning-based RRW scheme. DRRW uses an Integer Invertible Watermark Network (iIWN) to map integer data distributions invertibly, addressing conventional RRW limitations. Unlike traditional RRW, which needs distortion-specific designs, DRRW employs an encoder-noise layer-decoder framework for adaptive robustness via end-to-end training. In inference, cover image and watermark map to an overflowed stego image and latent variables, compressed by arithmetic coding into a bitstream embedded via reversible data hiding for lossless recovery. We introduce an overflow penalty loss to reduce pixel overflow, shortening the auxiliary bitstream while enhancing robustness and stego image quality. An adaptive weight adjustment strategy avoids manual watermark loss weighting, improving training stability and performance. Experiments show DRRW outperforms state-of-the-art RRW methods, boosting robustness and cutting embedding, extraction, and recovery complexities by 55.14\(\times\), 5.95\(\times\), and 3.57\(\times\), respectively. The auxiliary bitstream shrinks by 43.86\(\times\), with reversible embedding succeeding on 16,762 PASCAL VOC 2012 images, advancing practical RRW. DRRW exceeds irreversible robust watermarking in robustness and quality while maintaining reversibility.
☆ Scene-based nonuniformity correction with homography transformation
Due to their affordable, low mass, and small dimensions, uncooled microbolometer-based thermal focal plane arrays (UC-FPAs) are useful for long-wave infrared (LWIR)imaging applications. However, in outdoor conditions typical in agricultural remote sensing, cameras based on UC-FPAs may suffer from drift in offset and gain. To tackle the persistent drift, the system requires continuous calibration. Our goal in this study was to eliminate this requirement via a computational schema. In a former study, we estimated unknown gain and offset values and thermographic images of an object from a sequence of pairs of successive images taken at two different blur levels.In the current work, we took on a similar problem using a sequence of shifted images, with relative shifts caused by realistic drone hovering modeled by homography transformation. This places our work in the realm of scene-based nonuniformity correction problems. We show that an object's thermographic values, as well as gain and offset, can be jointly estimated by relying on a few sets of shifted images. We use a minimum likelihood estimator, which is found using alternating minimization. Registration is done using a generalized Lucas-Kanade method. Simulations show promising accuracy with mean Pearson correlation of more than 0.9999998 between ground truth and restoration. Under ideal assumptions, this is equivalent to a mean restoration error of less than 0.01 Celsius degree.
comment: Imaging, Inverse problems, Functional analysis, Blind deconvolution
☆ ERetinex: Event Camera Meets Retinex Theory for Low-Light Image Enhancement ICRA 2025
Low-light image enhancement aims to restore the under-exposure image captured in dark scenarios. Under such scenarios, traditional frame-based cameras may fail to capture the structure and color information due to the exposure time limitation. Event cameras are bio-inspired vision sensors that respond to pixel-wise brightness changes asynchronously. Event cameras' high dynamic range is pivotal for visual perception in extreme low-light scenarios, surpassing traditional cameras and enabling applications in challenging dark environments. In this paper, inspired by the success of the retinex theory for traditional frame-based low-light image restoration, we introduce the first methods that combine the retinex theory with event cameras and propose a novel retinex-based low-light image restoration framework named ERetinex. Among our contributions, the first is developing a new approach that leverages the high temporal resolution data from event cameras with traditional image information to estimate scene illumination accurately. This method outperforms traditional image-only techniques, especially in low-light environments, by providing more precise lighting information. Additionally, we propose an effective fusion strategy that combines the high dynamic range data from event cameras with the color information of traditional images to enhance image quality. Through this fusion, we can generate clearer and more detail-rich images, maintaining the integrity of visual information even under extreme lighting conditions. The experimental results indicate that our proposed method outperforms state-of-the-art (SOTA) methods, achieving a gain of 1.0613 dB in PSNR while reducing FLOPS by \textbf{84.28}\%.
comment: Accepted to ICRA 2025
☆ A Novel Streamline-based diffusion MRI Tractography Registration Method with Probabilistic Keypoint Detection
Registration of diffusion MRI tractography is an essential step for analyzing group similarities and variations in the brain's white matter (WM). Streamline-based registration approaches can leverage the 3D geometric information of fiber pathways to enable spatial alignment after registration. Existing methods usually rely on the optimization of the spatial distances to identify the optimal transformation. However, such methods overlook point connectivity patterns within the streamline itself, limiting their ability to identify anatomical correspondences across tractography datasets. In this work, we propose a novel unsupervised approach using deep learning to perform streamline-based dMRI tractography registration. The overall idea is to identify corresponding keypoint pairs across subjects for spatial alignment of tractography datasets. We model tractography as point clouds to leverage the graph connectivity along streamlines. We propose a novel keypoint detection method for streamlines, framed as a probabilistic classification task to identify anatomically consistent correspondences across unstructured streamline sets. In the experiments, we compare several existing methods and show highly effective and efficient tractography registration performance.
☆ BioD2C: A Dual-level Semantic Consistency Constraint Framework for Biomedical VQA
Biomedical visual question answering (VQA) has been widely studied and has demonstrated significant application value and potential in fields such as assistive medical diagnosis. Despite their success, current biomedical VQA models perform multimodal information interaction only at the model level within large language models (LLMs), leading to suboptimal multimodal semantic alignment when dealing with complex tasks. To address this issue, we propose BioD2C: a novel Dual-level Semantic Consistency Constraint Framework for Biomedical VQA, which achieves dual-level semantic interaction alignment at both the model and feature levels, enabling the model to adaptively learn visual features based on the question. Specifically, we firstly integrate textual features into visual features via an image-text fusion mechanism as feature-level semantic interaction, obtaining visual features conditioned on the given text; and then introduce a text-queue-based cross-modal soft semantic loss function to further align the image semantics with the question semantics. Specifically, in this work, we establish a new dataset, BioVGQ, to address inherent biases in prior datasets by filtering manually-altered images and aligning question-answer pairs with multimodal context, and train our model on this dataset. Extensive experimental results demonstrate that BioD2C achieves state-of-the-art (SOTA) performance across multiple downstream datasets, showcasing its robustness, generalizability, and potential to advance biomedical VQA research.
☆ Exploring Token-Level Augmentation in Vision Transformer for Semi-Supervised Semantic Segmentation
Semi-supervised semantic segmentation has witnessed remarkable advancements in recent years. However, existing algorithms are based on convolutional neural networks and directly applying them to Vision Transformers poses certain limitations due to conceptual disparities. To this end, we propose TokenMix, a data augmentation technique specifically designed for semi-supervised semantic segmentation with Vision Transformers. TokenMix aligns well with the global attention mechanism by mixing images at the token level, enhancing learning capability for contexutual information among image patches. We further incorporate image augmentation and feature augmentation to promote the diversity of augmentation. Moreover, to enhance consistency regularization, we propose a dual-branch framework where each branch applies both image augmentation and feature augmentation to the input image. We conduct extensive experiments across multiple benchmark datasets, including Pascal VOC 2012, Cityscapes, and COCO. Results suggest that the proposed method outperforms state-of-the-art algorithms with notably observed accuracy improvement, especially under the circumstance of limited fine annotations.
☆ 2DGS-Avatar: Animatable High-fidelity Clothed Avatar via 2D Gaussian Splatting
Real-time rendering of high-fidelity and animatable avatars from monocular videos remains a challenging problem in computer vision and graphics. Over the past few years, the Neural Radiance Field (NeRF) has made significant progress in rendering quality but behaves poorly in run-time performance due to the low efficiency of volumetric rendering. Recently, methods based on 3D Gaussian Splatting (3DGS) have shown great potential in fast training and real-time rendering. However, they still suffer from artifacts caused by inaccurate geometry. To address these problems, we propose 2DGS-Avatar, a novel approach based on 2D Gaussian Splatting (2DGS) for modeling animatable clothed avatars with high-fidelity and fast training performance. Given monocular RGB videos as input, our method generates an avatar that can be driven by poses and rendered in real-time. Compared to 3DGS-based methods, our 2DGS-Avatar retains the advantages of fast training and rendering while also capturing detailed, dynamic, and photo-realistic appearances. We conduct abundant experiments on popular datasets such as AvatarRex and THuman4.0, demonstrating impressive performance in both qualitative and quantitative metrics.
comment: ICVRV 2024
☆ Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection CVPR2025
Anomaly detection (AD) is essential for industrial inspection, yet existing methods typically rely on ``comparing'' test images to normal references from a training set. However, variations in appearance and positioning often complicate the alignment of these references with the test image, limiting detection accuracy. We observe that most anomalies manifest as local variations, meaning that even within anomalous images, valuable normal information remains. We argue that this information is useful and may be more aligned with the anomalies since both the anomalies and the normal information originate from the same image. Therefore, rather than relying on external normality from the training set, we propose INP-Former, a novel method that extracts Intrinsic Normal Prototypes (INPs) directly from the test image. Specifically, we introduce the INP Extractor, which linearly combines normal tokens to represent INPs. We further propose an INP Coherence Loss to ensure INPs can faithfully represent normality for the testing image. These INPs then guide the INP-Guided Decoder to reconstruct only normal tokens, with reconstruction errors serving as anomaly scores. Additionally, we propose a Soft Mining Loss to prioritize hard-to-optimize samples during training. INP-Former achieves state-of-the-art performance in single-class, multi-class, and few-shot AD tasks across MVTec-AD, VisA, and Real-IAD, positioning it as a versatile and universal solution for AD. Remarkably, INP-Former also demonstrates some zero-shot AD capability. Code is available at:https://github.com/luow23/INP-Former.
comment: Accepted by CVPR2025
☆ Exploring Model Quantization in GenAI-based Image Inpainting and Detection of Arable Plants
Deep learning-based weed control systems often suffer from limited training data diversity and constrained on-board computation, impacting their real-world performance. To overcome these challenges, we propose a framework that leverages Stable Diffusion-based inpainting to augment training data progressively in 10% increments -- up to an additional 200%, thus enhancing both the volume and diversity of samples. Our approach is evaluated on two state-of-the-art object detection models, YOLO11(l) and RT-DETR(l), using the mAP50 metric to assess detection performance. We explore quantization strategies (FP16 and INT8) for both the generative inpainting and detection models to strike a balance between inference speed and accuracy. Deployment of the downstream models on the Jetson Orin Nano demonstrates the practical viability of our framework in resource-constrained environments, ultimately improving detection accuracy and computational efficiency in intelligent weed management systems.
☆ InfoGNN: End-to-end deep learning on mesh via graph neural networks
3D models are widely used in various industries, and mesh data has become an indispensable part of 3D modeling because of its unique advantages. Mesh data can provide an intuitive and practical expression of rich 3D information. However, its disordered, irregular data structure and complex surface information make it challenging to apply with deep learning models directly. Traditional mesh data processing methods often rely on mesh models with many limitations, such as manifold, which restrict their application scopes in reality and do not fully utilize the advantages of mesh models. This paper proposes a novel end-to-end framework for addressing the challenges associated with deep learning in mesh models centered around graph neural networks (GNN) and is titled InfoGNN. InfoGNN treats the mesh model as a graph, which enables it to handle irregular mesh data efficiently. Moreover, we propose InfoConv and InfoMP modules, which utilize the position information of the points and fully use the static information such as face normals, dihedral angles, and dynamic global feature information to fully use all kinds of data. In addition, InfoGNN is an end-to-end framework, and we simplify the network design to make it more efficient, paving the way for efficient deep learning of complex 3D models. We conducted experiments on several publicly available datasets, and the results show that InfoGNN achieves excellent performance in mesh classification and segmentation tasks.
☆ Building 3D In-Context Learning Universal Model in Neuroimaging
In-context learning (ICL), a type of universal model, demonstrates exceptional generalization across a wide range of tasks without retraining by leveraging task-specific guidance from context, making it particularly effective for the complex demands of neuroimaging. However, existing ICL models, which take 2D images as input, struggle to fully leverage the 3D anatomical structures in neuroimages, leading to a lack of global awareness and suboptimal performance. In this regard, we introduce Neuroverse3D, an ICL model capable of performing multiple neuroimaging tasks (e.g., segmentation, denoising, inpainting) in 3D. Neuroverse3D overcomes the large memory consumption due to 3D inputs through adaptive parallel-sequential context processing and a U-shape fusion strategy, allowing it to handle an unlimited number of context images. Additionally, we propose an optimized loss to balance multi-task training and enhance the focus on anatomical structures. Our study incorporates 43,674 3D scans from 19 neuroimaging datasets and evaluates Neuroverse3D on 14 diverse tasks using held-out test sets. The results demonstrate that Neuroverse3D significantly outperforms existing ICL models and closely matches the performance of task-specific models. The code and model weights are publicly released at: https://github.com/jiesihu/Neu3D.
☆ VisAgent: Narrative-Preserving Story Visualization Framework ICASSP 2025
Story visualization is the transformation of narrative elements into image sequences. While existing research has primarily focused on visual contextual coherence, the deeper narrative essence of stories often remains overlooked. This limitation hinders the practical application of these approaches, as generated images frequently fail to capture the intended meaning and nuances of the narrative fully. To address these challenges, we propose VisAgent, a training-free multi-agent framework designed to comprehend and visualize pivotal scenes within a given story. By considering story distillation, semantic consistency, and contextual coherence, VisAgent employs an agentic workflow. In this workflow, multiple specialized agents collaborate to: (i) refine layered prompts based on the narrative structure and (ii) seamlessly integrate \gt{generated} elements, including refined prompts, scene elements, and subject placement, into the final image. The empirically validated effectiveness confirms the framework's suitability for practical story visualization applications.
comment: Accepted to ICASSP 2025. Equal contribution from first two authors
☆ BHViT: Binarized Hybrid Vision Transformer CVPR2025
Model binarization has made significant progress in enabling real-time and energy-efficient computation for convolutional neural networks (CNN), offering a potential solution to the deployment challenges faced by Vision Transformers (ViTs) on edge devices. However, due to the structural differences between CNN and Transformer architectures, simply applying binary CNN strategies to the ViT models will lead to a significant performance drop. To tackle this challenge, we propose BHViT, a binarization-friendly hybrid ViT architecture and its full binarization model with the guidance of three important observations. Initially, BHViT utilizes the local information interaction and hierarchical feature aggregation technique from coarse to fine levels to address redundant computations stemming from excessive tokens. Then, a novel module based on shift operations is proposed to enhance the performance of the binary Multilayer Perceptron (MLP) module without significantly increasing computational overhead. In addition, an innovative attention matrix binarization method based on quantization decomposition is proposed to evaluate the token's importance in the binarized attention matrix. Finally, we propose a regularization loss to address the inadequate optimization caused by the incompatibility between the weight oscillation in the binary layers and the Adam Optimizer. Extensive experimental results demonstrate that our proposed algorithm achieves SOTA performance among binary ViT methods.
comment: Accepted by CVPR2025
☆ Vision-Language Model IP Protection via Prompt-based Learning
Vision-language models (VLMs) like CLIP (Contrastive Language-Image Pre-Training) have seen remarkable success in visual recognition, highlighting the increasing need to safeguard the intellectual property (IP) of well-trained models. Effective IP protection extends beyond ensuring authorized usage; it also necessitates restricting model deployment to authorized data domains, particularly when the model is fine-tuned for specific target domains. However, current IP protection methods often rely solely on the visual backbone, which may lack sufficient semantic richness. To bridge this gap, we introduce IP-CLIP, a lightweight IP protection strategy tailored to CLIP, employing a prompt-based learning approach. By leveraging the frozen visual backbone of CLIP, we extract both image style and content information, incorporating them into the learning of IP prompt. This strategy acts as a robust barrier, effectively preventing the unauthorized transfer of features from authorized domains to unauthorized ones. Additionally, we propose a style-enhancement branch that constructs feature banks for both authorized and unauthorized domains. This branch integrates self-enhanced and cross-domain features, further strengthening IP-CLIP's capability to block features from unauthorized domains. Finally, we present new three metrics designed to better balance the performance degradation of authorized and unauthorized domains. Comprehensive experiments in various scenarios demonstrate its promising potential for application in IP protection tasks for VLMs.
☆ PIDLoc: Cross-View Pose Optimization Network Inspired by PID Controllers CVPR-25
Accurate localization is essential for autonomous driving, but GNSS-based methods struggle in challenging environments such as urban canyons. Cross-view pose optimization offers an effective solution by directly estimating vehicle pose using satellite-view images. However, existing methods primarily rely on cross-view features at a given pose, neglecting fine-grained contexts for precision and global contexts for robustness against large initial pose errors. To overcome these limitations, we propose PIDLoc, a novel cross-view pose optimization approach inspired by the proportional-integral-derivative (PID) controller. Using RGB images and LiDAR, the PIDLoc comprises the PID branches to model cross-view feature relationships and the spatially aware pose estimator (SPE) to estimate the pose from these relationships. The PID branches leverage feature differences for local context (P), aggregated feature differences for global context (I), and gradients of feature differences for precise pose adjustment (D) to enhance localization accuracy under large initial pose errors. Integrated with the PID branches, the SPE captures spatial relationships within the PID-branch features for consistent localization. Experimental results demonstrate that the PIDLoc achieves state-of-the-art performance in cross-view pose estimation for the KITTI dataset, reducing position error by $37.8\%$ compared with the previous state-of-the-art.
comment: Accepted by CVPR-25
☆ Teaching Metric Distance to Autoregressive Multimodal Foundational Models
As large language models expand beyond natural language to domains such as mathematics, multimodal understanding, and embodied agents, tokens increasingly reflect metric relationships rather than purely linguistic meaning. We introduce DIST2Loss, a distance-aware framework designed to train autoregressive discrete models by leveraging predefined distance relationships among output tokens. At its core, DIST2Loss transforms continuous exponential family distributions derived from inherent distance metrics into discrete, categorical optimization targets compatible with the models' architectures. This approach enables the models to learn and preserve meaningful distance relationships during token generation while maintaining compatibility with existing architectures. Empirical evaluations show consistent performance gains in diverse multimodal applications, including visual grounding, robotic manipulation, generative reward modeling, and image generation using vector-quantized features. These improvements are pronounced in cases of limited training data, highlighting DIST2Loss's effectiveness in resource-constrained settings.
☆ mmDEAR: mmWave Point Cloud Density Enhancement for Accurate Human Body Reconstruction
Millimeter-wave (mmWave) radar offers robust sensing capabilities in diverse environments, making it a highly promising solution for human body reconstruction due to its privacy-friendly and non-intrusive nature. However, the significant sparsity of mmWave point clouds limits the estimation accuracy. To overcome this challenge, we propose a two-stage deep learning framework that enhances mmWave point clouds and improves human body reconstruction accuracy. Our method includes a mmWave point cloud enhancement module that densifies the raw data by leveraging temporal features and a multi-stage completion network, followed by a 2D-3D fusion module that extracts both 2D and 3D motion features to refine SMPL parameters. The mmWave point cloud enhancement module learns the detailed shape and posture information from 2D human masks in single-view images. However, image-based supervision is involved only during the training phase, and the inference relies solely on sparse point clouds to maintain privacy. Experiments on multiple datasets demonstrate that our approach outperforms state-of-the-art methods, with the enhanced point clouds further improving performance when integrated into existing models.
☆ Label-Efficient LiDAR Panoptic Segmentation
A main bottleneck of learning-based robotic scene understanding methods is the heavy reliance on extensive annotated training data, which often limits their generalization ability. In LiDAR panoptic segmentation, this challenge becomes even more pronounced due to the need to simultaneously address both semantic and instance segmentation from complex, high-dimensional point cloud data. In this work, we address the challenge of LiDAR panoptic segmentation with very few labeled samples by leveraging recent advances in label-efficient vision panoptic segmentation. To this end, we propose a novel method, Limited-Label LiDAR Panoptic Segmentation (L3PS), which requires only a minimal amount of labeled data. Our approach first utilizes a label-efficient 2D network to generate panoptic pseudo-labels from a small set of annotated images, which are subsequently projected onto point clouds. We then introduce a novel 3D refinement module that capitalizes on the geometric properties of point clouds. By incorporating clustering techniques, sequential scan accumulation, and ground point separation, this module significantly enhances the accuracy of the pseudo-labels, improving segmentation quality by up to +10.6 PQ and +7.9 mIoU. We demonstrate that these refined pseudo-labels can be used to effectively train off-the-shelf LiDAR segmentation networks. Through extensive experiments, we show that L3PS not only outperforms existing methods but also substantially reduces the annotation burden. We release the code of our work at https://l3ps.cs.uni-freiburg.de.
☆ BdSLW401: Transformer-Based Word-Level Bangla Sign Language Recognition Using Relative Quantization Encoding (RQE)
Sign language recognition (SLR) for low-resource languages like Bangla suffers from signer variability, viewpoint variations, and limited annotated datasets. In this paper, we present BdSLW401, a large-scale, multi-view, word-level Bangla Sign Language (BdSL) dataset with 401 signs and 102,176 video samples from 18 signers in front and lateral views. To improve transformer-based SLR, we introduce Relative Quantization Encoding (RQE), a structured embedding approach anchoring landmarks to physiological reference points and quantize motion trajectories. RQE improves attention allocation by decreasing spatial variability, resulting in 44.3% WER reduction in WLASL100, 21.0% in SignBD-200, and significant gains in BdSLW60 and SignBD-90. However, fixed quantization becomes insufficient on large-scale datasets (e.g., WLASL2000), indicating the need for adaptive encoding strategies. Further, RQE-SF, an extended variant that stabilizes shoulder landmarks, achieves improvements in pose consistency at the cost of small trade-offs in lateral view recognition. The attention graphs prove that RQE improves model interpretability by focusing on the major articulatory features (fingers, wrists) and the more distinctive frames instead of global pose changes. Introducing BdSLW401 and demonstrating the effectiveness of RQE-enhanced structured embeddings, this work advances transformer-based SLR for low-resource languages and sets a benchmark for future research in this area.
☆ Are Large Vision Language Models Good Game Players? ICLR2025
Large Vision Language Models (LVLMs) have demonstrated remarkable abilities in understanding and reasoning about both visual and textual information. However, existing evaluation methods for LVLMs, primarily based on benchmarks like Visual Question Answering and image captioning, often fail to capture the full scope of LVLMs' capabilities. These benchmarks are limited by issues such as inadequate assessment of detailed visual perception, data contamination, and a lack of focus on multi-turn reasoning. To address these challenges, we propose \method{}, a game-based evaluation framework designed to provide a comprehensive assessment of LVLMs' cognitive and reasoning skills in structured environments. \method{} uses a set of games to evaluate LVLMs on four core tasks: Perceiving, Question Answering, Rule Following, and End-to-End Playing, with each target task designed to assess specific abilities, including visual perception, reasoning, decision-making, etc. Based on this framework, we conduct extensive experiments that explore the limitations of current LVLMs, such as handling long structured outputs and perceiving detailed and dense elements. Code and data are publicly available at https://github.com/xinke-wang/LVLM-Playground.
comment: ICLR2025
☆ Q-Eval-100K: Evaluating Visual Quality and Alignment Level for Text-to-Vision Content CVPR 2025
Evaluating text-to-vision content hinges on two crucial aspects: visual quality and alignment. While significant progress has been made in developing objective models to assess these dimensions, the performance of such models heavily relies on the scale and quality of human annotations. According to Scaling Law, increasing the number of human-labeled instances follows a predictable pattern that enhances the performance of evaluation models. Therefore, we introduce a comprehensive dataset designed to Evaluate Visual quality and Alignment Level for text-to-vision content (Q-EVAL-100K), featuring the largest collection of human-labeled Mean Opinion Scores (MOS) for the mentioned two aspects. The Q-EVAL-100K dataset encompasses both text-to-image and text-to-video models, with 960K human annotations specifically focused on visual quality and alignment for 100K instances (60K images and 40K videos). Leveraging this dataset with context prompt, we propose Q-Eval-Score, a unified model capable of evaluating both visual quality and alignment with special improvements for handling long-text prompt alignment. Experimental results indicate that the proposed Q-Eval-Score achieves superior performance on both visual quality and alignment, with strong generalization capabilities across other benchmarks. These findings highlight the significant value of the Q-EVAL-100K dataset. Data and codes will be available at https://github.com/zzc-1998/Q-Eval.
comment: Accepted to CVPR 2025
☆ YOLO-PRO: Enhancing Instance-Specific Object Detection with Full-Channel Global Self-Attention
This paper addresses the inherent limitations of conventional bottleneck structures (diminished instance discriminability due to overemphasis on batch statistics) and decoupled heads (computational redundancy) in object detection frameworks by proposing two novel modules: the Instance-Specific Bottleneck with full-channel global self-attention (ISB) and the Instance-Specific Asymmetric Decoupled Head (ISADH). The ISB module innovatively reconstructs feature maps to establish an efficient full-channel global attention mechanism through synergistic fusion of batch-statistical and instance-specific features. Complementing this, the ISADH module pioneers an asymmetric decoupled architecture enabling hierarchical multi-dimensional feature integration via dual-stream batch-instance representation fusion. Extensive experiments on the MS-COCO benchmark demonstrate that the coordinated deployment of ISB and ISADH in the YOLO-PRO framework achieves state-of-the-art performance across all computational scales. Specifically, YOLO-PRO surpasses YOLOv8 by 1.0-1.6% AP (N/S/M/L/X scales) and outperforms YOLO11 by 0.1-0.5% AP in critical M/L/X groups, while maintaining competitive computational efficiency. This work provides practical insights for developing high-precision detectors deployable on edge devices.
☆ CQ CNN: A Hybrid Classical Quantum Convolutional Neural Network for Alzheimer's Disease Detection Using Diffusion Generated and U Net Segmented 3D MRI
The detection of Alzheimer disease (AD) from clinical MRI data is an active area of research in medical imaging. Recent advances in quantum computing, particularly the integration of parameterized quantum circuits (PQCs) with classical machine learning architectures, offer new opportunities to develop models that may outperform traditional methods. However, quantum machine learning (QML) remains in its early stages and requires further experimental analysis to better understand its behavior and limitations. In this paper, we propose an end to end hybrid classical quantum convolutional neural network (CQ CNN) for AD detection using clinically formatted 3D MRI data. Our approach involves developing a framework to make 3D MRI data usable for machine learning, designing and training a brain tissue segmentation model (Skull Net), and training a diffusion model to generate synthetic images for the minority class. Our converged models exhibit potential quantum advantages, achieving higher accuracy in fewer epochs than classical models. The proposed beta8 3 qubit model achieves an accuracy of 97.50%, surpassing state of the art (SOTA) models while requiring significantly fewer computational resources. In particular, the architecture employs only 13K parameters (0.48 MB), reducing the parameter count by more than 99.99% compared to current SOTA models. Furthermore, the diffusion-generated data used to train our quantum models, in conjunction with real samples, preserve clinical structural standards, representing a notable first in the field of QML. We conclude that CQCNN architecture like models, with further improvements in gradient optimization techniques, could become a viable option and even a potential alternative to classical models for AD detection, especially in data limited and resource constrained clinical settings.
comment: Application of hybrid quantum-classical machine learning for (early stage) disease detection
☆ GRADEO: Towards Human-Like Evaluation for Text-to-Video Generation via Multi-Step Reasoning
Recent great advances in video generation models have demonstrated their potential to produce high-quality videos, bringing challenges to effective evaluation. Unlike human evaluation, existing automated evaluation metrics lack high-level semantic understanding and reasoning capabilities for video, thus making them infeasible and unexplainable. To fill this gap, we curate GRADEO-Instruct, a multi-dimensional T2V evaluation instruction tuning dataset, including 3.3k videos from over 10 existing video generation models and multi-step reasoning assessments converted by 16k human annotations. We then introduce GRADEO, one of the first specifically designed video evaluation models, which grades AI-generated videos for explainable scores and assessments through multi-step reasoning. Experiments show that our method aligns better with human evaluations than existing methods. Furthermore, our benchmarking reveals that current video generation models struggle to produce content that aligns with human reasoning and complex real-world scenarios. The models, datasets, and codes will be released soon.
☆ BiasICL: In-Context Learning and Demographic Biases of Vision Language Models
Vision language models (VLMs) show promise in medical diagnosis, but their performance across demographic subgroups when using in-context learning (ICL) remains poorly understood. We examine how the demographic composition of demonstration examples affects VLM performance in two medical imaging tasks: skin lesion malignancy prediction and pneumothorax detection from chest radiographs. Our analysis reveals that ICL influences model predictions through multiple mechanisms: (1) ICL allows VLMs to learn subgroup-specific disease base rates from prompts and (2) ICL leads VLMs to make predictions that perform differently across demographic groups, even after controlling for subgroup-specific disease base rates. Our empirical results inform best-practices for prompting current VLMs (specifically examining demographic subgroup performance, and matching base rates of labels to target distribution at a bulk level and within subgroups), while also suggesting next steps for improving our theoretical understanding of these models.
☆ COMMA: Coordinate-aware Modulated Mamba Network for 3D Dispersed Vessel Segmentation
Accurate segmentation of 3D vascular structures is essential for various medical imaging applications. The dispersed nature of vascular structures leads to inherent spatial uncertainty and necessitates location awareness, yet most current 3D medical segmentation models rely on the patch-wise training strategy that usually loses this spatial context. In this study, we introduce the Coordinate-aware Modulated Mamba Network (COMMA) and contribute a manually labeled dataset of 570 cases, the largest publicly available 3D vessel dataset to date. COMMA leverages both entire and cropped patch data through global and local branches, ensuring robust and efficient spatial location awareness. Specifically, COMMA employs a channel-compressed Mamba (ccMamba) block to encode entire image data, capturing long-range dependencies while optimizing computational costs. Additionally, we propose a coordinate-aware modulated (CaM) block to enhance interactions between the global and local branches, allowing the local branch to better perceive spatial information. We evaluate COMMA on six datasets, covering two imaging modalities and five types of vascular tissues. The results demonstrate COMMA's superior performance compared to state-of-the-art methods with computational efficiency, especially in segmenting small vessels. Ablation studies further highlight the importance of our proposed modules and spatial information. The code and data will be open source at https://github.com/shigen-StoneRoot/COMMA.
☆ Exploring Simple Siamese Network for High-Resolution Video Quality Assessment ICASSP 2025
In the research of video quality assessment (VQA), two-branch network has emerged as a promising solution. It decouples VQA with separate technical and aesthetic branches to measure the perception of low-level distortions and high-level semantics respectively. However, we argue that while technical and aesthetic perspectives are complementary, the technical perspective itself should be measured in semantic-aware manner. We hypothesize that existing technical branch struggles to perceive the semantics of high-resolution videos, as it is trained on local mini-patches sampled from videos. This issue can be hidden by apparently good results on low-resolution videos, but indeed becomes critical for high-resolution VQA. This work introduces SiamVQA, a simple but effective Siamese network for highre-solution VQA. SiamVQA shares weights between technical and aesthetic branches, enhancing the semantic perception ability of technical branch to facilitate technical-quality representation learning. Furthermore, it integrates a dual cross-attention layer for fusing technical and aesthetic features. SiamVQA achieves state-of-the-art accuracy on high-resolution benchmarks, and competitive results on lower-resolution benchmarks. Codes will be available at: https://github.com/srcn-ivl/SiamVQA
comment: Accepted by ICASSP 2025
☆ Generative Model-Assisted Demosaicing for Cross-multispectral Cameras
As a crucial part of the spectral filter array (SFA)-based multispectral imaging process, spectral demosaicing has exploded with the proliferation of deep learning techniques. However, (1) bothering by the difficulty of capturing corresponding labels for real data or simulating the practical spectral imaging process, end-to-end networks trained in a supervised manner using simulated data often perform poorly on real data. (2) cross-camera spectral discrepancies make it difficult to apply pre-trained models to new cameras. (3) existing demosaicing networks are prone to introducing visual artifacts on hard cases due to the interpolation of unknown values. To address these issues, we propose a hybrid supervised training method with the assistance of the self-supervised generative model, which performs well on real data across different spectral cameras. Specifically, our approach consists of three steps: (1) Pre-Training step: training the end-to-end neural network on a large amount of simulated data; (2) Pseudo-Pairing step: generating pseudo-labels of real target data using the self-supervised generative model; (3) Fine-Tuning step: fine-tuning the pre-trained model on the pseudo data pairs obtained in (2). To alleviate artifacts, we propose a frequency-domain hard patch selection method that identifies artifact-prone regions by analyzing spectral discrepancies using Fourier transform and filtering techniques, allowing targeted fine-tuning to enhance demosaicing performance. Finally, we propose UniSpecTest, a real-world multispectral mosaic image dataset for testing. Ablation experiments have demonstrated the effectiveness of each training step, and extensive experiments on both synthetic and real datasets show that our method achieves significant performance gains compared to state-of-the-art techniques.
☆ Semantic Prior Distillation with Vision Foundation Model for Enhanced Rapid Bone Scintigraphy Image Restoration
Rapid bone scintigraphy is an essential tool for diagnosing skeletal diseases and tumor metastasis in pediatric patients, as it reduces scan time and minimizes patient discomfort. However, rapid scans often result in poor image quality, potentially affecting diagnosis due to reduced resolution and detail, which make it challenging to identify and evaluate finer anatomical structures. To address this issue, we propose the first application of SAM-based semantic priors for medical image restoration, leveraging the Segment Anything Model (SAM) to enhance rapid bone scintigraphy images in pediatric populations. Our method comprises two cascaded networks, $f^{IR1}$ and $f^{IR2}$, augmented by three key modules: a Semantic Prior Integration (SPI) module, a Semantic Knowledge Distillation (SKD) module, and a Semantic Consistency Module (SCM). The SPI and SKD modules incorporate domain-specific semantic information from a fine-tuned SAM, while the SCM maintains consistent semantic feature representation throughout the cascaded networks. In addition, we will release a novel Rapid Bone Scintigraphy dataset called RBS, the first dataset dedicated to rapid bone scintigraphy image restoration in pediatric patients. RBS consists of 137 pediatric patients aged between 0.5 and 16 years who underwent both standard and rapid bone scans. The dataset includes scans performed at 20 cm/min (standard) and 40 cm/min (rapid), representing a $2\times$ acceleration. We conducted extensive experiments on both the publicly available endoscopic dataset and RBS. The results demonstrate that our method outperforms all existing methods across various metrics, including PSNR, SSIM, FID, and LPIPS.
comment: 12 pages, 9 figures, 8 tables
☆ Unified Arbitrary-Time Video Frame Interpolation and Prediction ICASSP 2025
Video frame interpolation and prediction aim to synthesize frames in-between and subsequent to existing frames, respectively. Despite being closely-related, these two tasks are traditionally studied with different model architectures, or same architecture but individually trained weights. Furthermore, while arbitrary-time interpolation has been extensively studied, the value of arbitrary-time prediction has been largely overlooked. In this work, we present uniVIP - unified arbitrary-time Video Interpolation and Prediction. Technically, we firstly extend an interpolation-only network for arbitrary-time interpolation and prediction, with a special input channel for task (interpolation or prediction) encoding. Then, we show how to train a unified model on common triplet frames. Our uniVIP provides competitive results for video interpolation, and outperforms existing state-of-the-arts for video prediction. Codes will be available at: https://github.com/srcn-ivl/uniVIP
comment: Accepted by ICASSP 2025
☆ A Token-level Text Image Foundation Model for Document Understanding
In recent years, general visual foundation models (VFMs) have witnessed increasing adoption, particularly as image encoders for popular multi-modal large language models (MLLMs). However, without semantically fine-grained supervision, these models still encounter fundamental prediction errors in the context of downstream text-image-related tasks, i.e., perception, understanding and reasoning with images containing small and dense texts. To bridge this gap, we develop TokenOCR, the first token-level visual foundation model specifically tailored for text-image-related tasks, designed to support a variety of traditional downstream applications. To facilitate the pretraining of TokenOCR, we also devise a high-quality data production pipeline that constructs the first token-level image text dataset, TokenIT, comprising 20 million images and 1.8 billion token-mask pairs. Furthermore, leveraging this foundation with exceptional image-as-text capability, we seamlessly replace previous VFMs with TokenOCR to construct a document-level MLLM, TokenVL, for VQA-based document understanding tasks. Finally, extensive experiments demonstrate the effectiveness of TokenOCR and TokenVL. Code, datasets, and weights will be available at https://token-family.github.io/TokenOCR_project.
comment: 23 pages
☆ On the Relationship Between Double Descent of CNNs and Shape/Texture Bias Under Learning Process
The double descent phenomenon, which deviates from the traditional bias-variance trade-off theory, attracts considerable research attention; however, the mechanism of its occurrence is not fully understood. On the other hand, in the study of convolutional neural networks (CNNs) for image recognition, methods are proposed to quantify the bias on shape features versus texture features in images, determining which features the CNN focuses on more. In this work, we hypothesize that there is a relationship between the shape/texture bias in the learning process of CNNs and epoch-wise double descent, and we conduct verification. As a result, we discover double descent/ascent of shape/texture bias synchronized with double descent of test error under conditions where epoch-wise double descent is observed. Quantitative evaluations confirm this correlation between the test errors and the bias values from the initial decrease to the full increase in test error. Interestingly, double descent/ascent of shape/texture bias is observed in some cases even in conditions without label noise, where double descent is thought not to occur. These experimental results are considered to contribute to the understanding of the mechanisms behind the double descent phenomenon and the learning process of CNNs in image recognition.
☆ Semi-Supervised Audio-Visual Video Action Recognition with Audio Source Localization Guided Mixup
Video action recognition is a challenging but important task for understanding and discovering what the video does. However, acquiring annotations for a video is costly, and semi-supervised learning (SSL) has been studied to improve performance even with a small number of labeled data in the task. Prior studies for semi-supervised video action recognition have mostly focused on using single modality - visuals - but the video is multi-modal, so utilizing both visuals and audio would be desirable and improve performance further, which has not been explored well. Therefore, we propose audio-visual SSL for video action recognition, which uses both visual and audio together, even with quite a few labeled data, which is challenging. In addition, to maximize the information of audio and video, we propose a novel audio source localization-guided mixup method that considers inter-modal relations between video and audio modalities. In experiments on UCF-51, Kinetics-400, and VGGSound datasets, our model shows the superior performance of the proposed semi-supervised audio-visual action recognition framework and audio source localization-guided mixup.
☆ SSNet: Saliency Prior and State Space Model-based Network for Salient Object Detection in RGB-D Images
Salient object detection (SOD) in RGB-D images is an essential task in computer vision, enabling applications in scene understanding, robotics, and augmented reality. However, existing methods struggle to capture global dependency across modalities, lack comprehensive saliency priors from both RGB and depth data, and are ineffective in handling low-quality depth maps. To address these challenges, we propose SSNet, a saliency-prior and state space model (SSM)-based network for the RGB-D SOD task. Unlike existing convolution- or transformer-based approaches, SSNet introduces an SSM-based multi-modal multi-scale decoder module to efficiently capture both intra- and inter-modal global dependency with linear complexity. Specifically, we propose a cross-modal selective scan SSM (CM-S6) mechanism, which effectively captures global dependency between different modalities. Furthermore, we introduce a saliency enhancement module (SEM) that integrates three saliency priors with deep features to refine feature representation and improve the localization of salient objects. To further address the issue of low-quality depth maps, we propose an adaptive contrast enhancement technique that dynamically refines depth maps, making them more suitable for the RGB-D SOD task. Extensive quantitative and qualitative experiments on seven benchmark datasets demonstrate that SSNet outperforms state-of-the-art methods.
♻ ☆ 3D-AffordanceLLM: Harnessing Large Language Models for Open-Vocabulary Affordance Detection in 3D Worlds ICLR
3D Affordance detection is a challenging problem with broad applications on various robotic tasks. Existing methods typically formulate the detection paradigm as a label-based semantic segmentation task. This paradigm relies on predefined labels and lacks the ability to comprehend complex natural language, resulting in limited generalization in open-world scene. To address these limitations, we reformulate the traditional affordance detection paradigm into \textit{Instruction Reasoning Affordance Segmentation} (IRAS) task. This task is designed to output a affordance mask region given a query reasoning text, which avoids fixed categories of input labels. We accordingly propose the \textit{3D-AffordanceLLM} (3D-ADLLM), a framework designed for reasoning affordance detection in 3D open-scene. Specifically, 3D-ADLLM introduces large language models (LLMs) to 3D affordance perception with a custom-designed decoder for generating affordance masks, thus achieving open-world reasoning affordance detection. In addition, given the scarcity of 3D affordance datasets for training large models, we seek to extract knowledge from general segmentation data and transfer it to affordance detection. Thus, we propose a multi-stage training strategy that begins with a novel pre-training task, i.e., \textit{Referring Object Part Segmentation}~(ROPS). This stage is designed to equip the model with general recognition and segmentation capabilities at the object-part level. Then followed by fine-tuning with the IRAS task, 3D-ADLLM obtains the reasoning ability for affordance detection. In summary, 3D-ADLLM leverages the rich world knowledge and human-object interaction reasoning ability of LLMs, achieving approximately an 8\% improvement in mIoU on open-vocabulary affordance detection tasks.
comment: ICLR
♻ ☆ A Survey on Vision-Language-Action Models for Embodied AI
Embodied AI is widely recognized as a key element of artificial general intelligence because it involves controlling embodied agents to perform tasks in the physical world. Building on the success of large language models and vision-language models, a new category of multimodal models -- referred to as vision-language-action models (VLAs) -- has emerged to address language-conditioned robotic tasks in embodied AI by leveraging their distinct ability to generate actions. In recent years, a myriad of VLAs have been developed, making it imperative to capture the rapidly evolving landscape through a comprehensive survey. To this end, we present the first survey on VLAs for embodied AI. This work provides a detailed taxonomy of VLAs, organized into three major lines of research. The first line focuses on individual components of VLAs. The second line is dedicated to developing control policies adept at predicting low-level actions. The third line comprises high-level task planners capable of decomposing long-horizon tasks into a sequence of subtasks, thereby guiding VLAs to follow more general user instructions. Furthermore, we provide an extensive summary of relevant resources, including datasets, simulators, and benchmarks. Finally, we discuss the challenges faced by VLAs and outline promising future directions in embodied AI. We have created a project associated with this survey, which is available at https://github.com/yueen-ma/Awesome-VLA.
comment: Project page: https://github.com/yueen-ma/Awesome-VLA
♻ ☆ WalnutData: A UAV Remote Sensing Dataset of Green Walnuts and Model Evaluation
The UAV technology is gradually maturing and can provide extremely powerful support for smart agriculture and precise monitoring. Currently, there is no dataset related to green walnuts in the field of agricultural computer vision. Thus, in order to promote the algorithm design in the field of agricultural computer vision, we used UAV to collect remote-sensing data from 8 walnut sample plots. Considering that green walnuts are subject to various lighting conditions and occlusion, we constructed a large-scale dataset with a higher-granularity of target features - WalnutData. This dataset contains a total of 30,240 images and 706,208 instances, and there are 4 target categories: being illuminated by frontal light and unoccluded (A1), being backlit and unoccluded (A2), being illuminated by frontal light and occluded (B1), and being backlit and occluded (B2). Subsequently, we evaluated many mainstream algorithms on WalnutData and used these evaluation results as the baseline standard. The dataset and all evaluation results can be obtained at https://github.com/1wuming/WalnutData.
♻ ☆ VILA-M3: Enhancing Vision-Language Models with Medical Expert Knowledge
Generalist vision language models (VLMs) have made significant strides in computer vision, but they fall short in specialized fields like healthcare, where expert knowledge is essential. In traditional computer vision tasks, creative or approximate answers may be acceptable, but in healthcare, precision is paramount.Current large multimodal models like Gemini and GPT-4o are insufficient for medical tasks due to their reliance on memorized internet knowledge rather than the nuanced expertise required in healthcare. VLMs are usually trained in three stages: vision pre-training, vision-language pre-training, and instruction fine-tuning (IFT). IFT has been typically applied using a mixture of generic and healthcare data. In contrast, we propose that for medical VLMs, a fourth stage of specialized IFT is necessary, which focuses on medical data and includes information from domain expert models. Domain expert models developed for medical use are crucial because they are specifically trained for certain clinical tasks, e.g. to detect tumors and classify abnormalities through segmentation and classification, which learn fine-grained features of medical data$-$features that are often too intricate for a VLM to capture effectively especially in radiology. This paper introduces a new framework, VILA-M3, for medical VLMs that utilizes domain knowledge via expert models. Through our experiments, we show an improved state-of-the-art (SOTA) performance with an average improvement of ~9% over the prior SOTA model Med-Gemini and ~6% over models trained on the specific tasks. Our approach emphasizes the importance of domain expertise in creating precise, reliable VLMs for medical applications.
♻ ☆ VOILA: Evaluation of MLLMs For Perceptual Understanding and Analogical Reasoning ICLR 2025
Multimodal Large Language Models (MLLMs) have become a powerful tool for integrating visual and textual information. Despite their exceptional performance on visual understanding benchmarks, measuring their ability to reason abstractly across multiple images remains a significant challenge. To address this, we introduce VOILA, a large-scale, open-ended, dynamic benchmark designed to evaluate MLLMs' perceptual understanding and abstract relational reasoning. VOILA employs an analogical mapping approach in the visual domain, requiring models to generate an image that completes an analogy between two given image pairs, reference and application, without relying on predefined choices. Our experiments demonstrate that the analogical reasoning tasks in VOILA present a challenge to MLLMs. Through multi-step analysis, we reveal that current MLLMs struggle to comprehend inter-image relationships and exhibit limited capabilities in high-level relational reasoning. Notably, we observe that performance improves when following a multi-step strategy of least-to-most prompting. Comprehensive evaluations on open-source models and GPT-4o show that on text-based answers, the best accuracy for challenging scenarios is 13% (LLaMa 3.2) and even for simpler tasks is only 29% (GPT-4o), while human performance is significantly higher at 70% across both difficulty levels.
comment: Accepted at ICLR 2025. Code and data: https://github.com/nlylmz/Voila
♻ ☆ CalibRefine: Deep Learning-Based Online Automatic Targetless LiDAR-Camera Calibration with Iterative and Attention-Driven Post-Refinement
Accurate multi-sensor calibration is essential for deploying robust perception systems in applications such as autonomous driving, robotics, and intelligent transportation. Existing LiDAR-camera calibration methods often rely on manually placed targets, preliminary parameter estimates, or intensive data preprocessing, limiting their scalability and adaptability in real-world settings. In this work, we propose a fully automatic, targetless, and online calibration framework, CalibRefine, which directly processes raw LiDAR point clouds and camera images. Our approach is divided into four stages: (1) a Common Feature Discriminator that trains on automatically detected objects--using relative positions, appearance embeddings, and semantic classes--to generate reliable LiDAR-camera correspondences, (2) a coarse homography-based calibration, (3) an iterative refinement to incrementally improve alignment as additional data frames become available, and (4) an attention-based refinement that addresses non-planar distortions by leveraging a Vision Transformer and cross-attention mechanisms. Through extensive experiments on two urban traffic datasets, we show that CalibRefine delivers high-precision calibration results with minimal human involvement, outperforming state-of-the-art targetless methods and remaining competitive with, or surpassing, manually tuned baselines. Our findings highlight how robust object-level feature matching, together with iterative and self-supervised attention-based adjustments, enables consistent sensor fusion in complex, real-world conditions without requiring ground-truth calibration matrices or elaborate data preprocessing.
comment: Submitted to Transportation Research Part C: Emerging Technologies
♻ ☆ AI-based association analysis for medical imaging using latent-space geometric confounder correction
This study addresses the challenges of confounding effects and interpretability in artificial-intelligence-based medical image analysis. Whereas existing literature often resolves confounding by removing confounder-related information from latent representations, this strategy risks affecting image reconstruction quality in generative models, thus limiting their applicability in feature visualization. To tackle this, we propose a different strategy that retains confounder-related information in latent representations while finding an alternative confounder-free representation of the image data. Our approach views the latent space of an autoencoder as a vector space, where imaging-related variables, such as the learning target (t) and confounder (c), have a vector capturing their variability. The confounding problem is addressed by searching a confounder-free vector which is orthogonal to the confounder-related vector but maximally collinear to the target-related vector. To achieve this, we introduce a novel correlation-based loss that not only performs vector searching in the latent space, but also encourages the encoder to generate latent representations linearly correlated with the variables. Subsequently, we interpret the confounder-free representation by sampling and reconstructing images along the confounder-free vector. The efficacy and flexibility of our proposed method are demonstrated across three applications, accommodating multiple confounders and utilizing diverse image modalities. Results affirm the method's effectiveness in reducing confounder influences, preventing wrong or misleading associations, and offering a unique visual interpretation for in-depth investigations by clinical and epidemiological researchers. The code is released in the following GitLab repository: https://gitlab.com/radiology/compopbio/ai_based_association_analysis}
comment: Accepted by Medical Image Analysis
♻ ☆ EC-DIT: Scaling Diffusion Transformers with Adaptive Expert-Choice Routing
Diffusion transformers have been widely adopted for text-to-image synthesis. While scaling these models up to billions of parameters shows promise, the effectiveness of scaling beyond current sizes remains underexplored and challenging. By explicitly exploiting the computational heterogeneity of image generations, we develop a new family of Mixture-of-Experts (MoE) models (EC-DIT) for diffusion transformers with expert-choice routing. EC-DIT learns to adaptively optimize the compute allocated to understand the input texts and generate the respective image patches, enabling heterogeneous computation aligned with varying text-image complexities. This heterogeneity provides an efficient way of scaling EC-DIT up to 97 billion parameters and achieving significant improvements in training convergence, text-to-image alignment, and overall generation quality over dense models and conventional MoE models. Through extensive ablations, we show that EC-DIT demonstrates superior scalability and adaptive compute allocation by recognizing varying textual importance through end-to-end training. Notably, in text-to-image alignment evaluation, our largest models achieve a state-of-the-art GenEval score of 71.68% and still maintain competitive inference speed with intuitive interpretability.
♻ ☆ Robust Long-Range Perception Against Sensor Misalignment in Autonomous Vehicles
Advances in machine learning algorithms for sensor fusion have significantly improved the detection and prediction of other road users, thereby enhancing safety. However, even a small angular displacement in the sensor's placement can cause significant degradation in output, especially at long range. In this paper, we demonstrate a simple yet generic and efficient multi-task learning approach that not only detects misalignment between different sensor modalities but is also robust against them for long-range perception. Along with the amount of misalignment, our method also predicts calibrated uncertainty, which can be useful for filtering and fusing predicted misalignment values over time. In addition, we show that the predicted misalignment parameters can be used for self-correcting input sensor data, further improving the perception performance under sensor misalignment.
♻ ☆ Efficient Connectivity-Preserving Instance Segmentation with Supervoxel-Based Loss Function
Reconstructing the intricate local morphology of neurons and their long-range projecting axons can address many connectivity related questions in neuroscience. The main bottleneck in connectomics pipelines is correcting topological errors, as multiple entangled neuronal arbors is a challenging instance segmentation problem. More broadly, segmentation of curvilinear, filamentous structures continues to pose significant challenges. To address this problem, we extend the notion of simple points from digital topology to connected sets of voxels (i.e. supervoxels) and propose a topology-aware neural network segmentation method with minimal computational overhead. We demonstrate its effectiveness on a new public dataset of 3-d light microscopy images of mouse brains, along with the benchmark datasets DRIVE, ISBI12, and CrackTree.
♻ ☆ Exploring Advanced Techniques for Visual Question Answering: A Comprehensive Comparison
Visual Question Answering (VQA) has emerged as a pivotal task in the intersection of computer vision and natural language processing, requiring models to understand and reason about visual content in response to natural language questions. Analyzing VQA datasets is essential for developing robust models that can handle the complexities of multimodal reasoning. Several approaches have been developed to examine these datasets, each offering distinct perspectives on question diversity, answer distribution, and visual-textual correlations. Despite significant progress, existing VQA models face challenges related to dataset bias, limited model complexity, commonsense reasoning gaps, rigid evaluation methods, and generalization to real world scenarios. This paper offers a detailed study of the original VQA dataset, baseline models and methods along with a comparative study of five advanced VQA models, ABC-CNN, KICNLE, Masked Vision and Language Modeling, BLIP-2, and OFA, each employing distinct methods to address these ongoing challenges.
comment: 8 pages, No figures
♻ ☆ VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language models (VLMs) use separate modules for understanding and generating visual content, which can lead to misalignment and increased complexity. In contrast, VILA-U employs a single autoregressive next-token prediction framework for both tasks, eliminating the need for additional components like diffusion models. This approach not only simplifies the model but also achieves near state-of-the-art performance in visual language understanding and generation. The success of VILA-U is attributed to two main factors: the unified vision tower that aligns discrete visual tokens with textual inputs during pretraining, which enhances visual perception, and autoregressive image generation can achieve similar quality as diffusion models with high-quality dataset. This allows VILA-U to perform comparably to more complex models using a fully token-based autoregressive framework.
comment: Code: https://github.com/mit-han-lab/vila-u. The first two authors contributed equally to this work
♻ ☆ CrowdSplat: Exploring Gaussian Splatting For Crowd Rendering
We present CrowdSplat, a novel approach that leverages 3D Gaussian Splatting for real-time, high-quality crowd rendering. Our method utilizes 3D Gaussian functions to represent animated human characters in diverse poses and outfits, which are extracted from monocular videos. We integrate Level of Detail (LoD) rendering to optimize computational efficiency and quality. The CrowdSplat framework consists of two stages: (1) avatar reconstruction and (2) crowd synthesis. The framework is also optimized for GPU memory usage to enhance scalability. Quantitative and qualitative evaluations show that CrowdSplat achieves good levels of rendering quality, memory efficiency, and computational performance. Through these experiments, we demonstrate that CrowdSplat is a viable solution for dynamic, realistic crowd simulation in real-time applications.
comment: 4 pages, 4 figures
♻ ☆ Evaluating CrowdSplat: Perceived Level of Detail for Gaussian Crowds
Efficient and realistic crowd rendering is an important element of many real-time graphics applications such as Virtual Reality (VR) and games. To this end, Levels of Detail (LOD) avatar representations such as polygonal meshes, image-based impostors, and point clouds have been proposed and evaluated. More recently, 3D Gaussian Splatting has been explored as a potential method for real-time crowd rendering. In this paper, we present a two-alternative forced choice (2AFC) experiment that aims to determine the perceived quality of 3D Gaussian avatars. Three factors were explored: Motion, LOD (i.e., #Gaussians), and the avatar height in Pixels (corresponding to the viewing distance). Participants viewed pairs of animated 3D Gaussian avatars and were tasked with choosing the most detailed one. Our findings can inform the optimization of LOD strategies in Gaussian-based crowd rendering, thereby helping to achieve efficient rendering while maintaining visual quality in real-time applications.
comment: 5 pages, 5 figures
♻ ☆ Near-infrared Image Deblurring and Event Denoising with Synergistic Neuromorphic Imaging
The fields of imaging in the nighttime dynamic and other extremely dark conditions have seen impressive and transformative advancements in recent years, partly driven by the rise of novel sensing approaches, e.g., near-infrared (NIR) cameras with high sensitivity and event cameras with minimal blur. However, inappropriate exposure ratios of near-infrared cameras make them susceptible to distortion and blur. Event cameras are also highly sensitive to weak signals at night yet prone to interference, often generating substantial noise and significantly degrading observations and analysis. Herein, we develop a new framework for low-light imaging combined with NIR imaging and event-based techniques, named synergistic neuromorphic imaging, which can jointly achieve NIR image deblurring and event denoising. Harnessing cross-modal features of NIR images and visible events via spectral consistency and higher-order interaction, the NIR images and events are simultaneously fused, enhanced, and bootstrapped. Experiments on real and realistically simulated sequences demonstrate the effectiveness of our method and indicate better accuracy and robustness than other methods in practical scenarios. This study gives impetus to enhance both NIR images and events, which paves the way for high-fidelity low-light imaging and neuromorphic reasoning.
♻ ☆ UFO: A Unified Approach to Fine-grained Visual Perception via Open-ended Language Interface
Generalist models have achieved remarkable success in both language and vision-language tasks, showcasing the potential of unified modeling. However, effectively integrating fine-grained perception tasks like detection and segmentation into these models remains a significant challenge. This is primarily because these tasks often rely heavily on task-specific designs and architectures that can complicate the modeling process. To address this challenge, we present \ours, a framework that \textbf{U}nifies \textbf{F}ine-grained visual perception tasks through an \textbf{O}pen-ended language interface. By transforming all perception targets into the language space, \ours unifies object-level detection, pixel-level segmentation, and image-level vision-language tasks into a single model. Additionally, we introduce a novel embedding retrieval approach that relies solely on the language interface to support segmentation tasks. Our framework bridges the gap between fine-grained perception and vision-language tasks, significantly simplifying architectural design and training strategies while achieving comparable or superior performance to methods with intricate task-specific designs. After multi-task training on five standard visual perception datasets, \ours outperforms the previous state-of-the-art generalist models by 12.3 mAP on COCO instance segmentation and 3.3 mIoU on ADE20K semantic segmentation. Furthermore, our method seamlessly integrates with existing MLLMs, effectively combining fine-grained perception capabilities with their advanced language abilities, thereby enabling more challenging tasks such as reasoning segmentation. Code and models are available at https://github.com/nnnth/UFO.
♻ ☆ HarmonySet: A Comprehensive Dataset for Understanding Video-Music Semantic Alignment and Temporal Synchronization CVPR 2025
This paper introduces HarmonySet, a comprehensive dataset designed to advance video-music understanding. HarmonySet consists of 48,328 diverse video-music pairs, annotated with detailed information on rhythmic synchronization, emotional alignment, thematic coherence, and cultural relevance. We propose a multi-step human-machine collaborative framework for efficient annotation, combining human insights with machine-generated descriptions to identify key transitions and assess alignment across multiple dimensions. Additionally, we introduce a novel evaluation framework with tasks and metrics to assess the multi-dimensional alignment of video and music, including rhythm, emotion, theme, and cultural context. Our extensive experiments demonstrate that HarmonySet, along with the proposed evaluation framework, significantly improves the ability of multimodal models to capture and analyze the intricate relationships between video and music.
comment: Accepted at CVPR 2025. Project page: https://harmonyset.github.io/
♻ ☆ A Comprehensive Survey on Composed Image Retrieval
Composed Image Retrieval (CIR) is an emerging yet challenging task that allows users to search for target images using a multimodal query, comprising a reference image and a modification text specifying the user's desired changes to the reference image. Given its significant academic and practical value, CIR has become a rapidly growing area of interest in the computer vision and machine learning communities, particularly with the advances in deep learning. To the best of our knowledge, there is currently no comprehensive review of CIR to provide a timely overview of this field. Therefore, we synthesize insights from over 120 publications in top conferences and journals, including ACM TOIS, SIGIR, and CVPR In particular, we systematically categorize existing supervised CIR and zero-shot CIR models using a fine-grained taxonomy. For a comprehensive review, we also briefly discuss approaches for tasks closely related to CIR, such as attribute-based CIR and dialog-based CIR. Additionally, we summarize benchmark datasets for evaluation and analyze existing supervised and zero-shot CIR methods by comparing experimental results across multiple datasets. Furthermore, we present promising future directions in this field, offering practical insights for researchers interested in further exploration. The curated collection of related works is maintained and continuously updated in https://github.com/haokunwen/Awesome-Composed-Image-Retrieval.
♻ ☆ WalkVLM:Aid Visually Impaired People Walking by Vision Language Model
Approximately 200 million individuals around the world suffer from varying degrees of visual impairment, making it crucial to leverage AI technology to offer walking assistance for these people. With the recent progress of vision-language models (VLMs), applying VLMs to offer walking guidance has become popular. However, the existing methods of walking guidance are mainly based on self-curated question-answering datasets that are not publicly accessible, without a standardized benchmark for training or evaluation. Moreover, walking assistance often requires real-time streaming video analysis and the generation of concise yet informative reminders, making VLMs struggle due to excessive responses and low efficiency in inferences. In this paper, we introduce the first large-scale dataset dedicated to walking assistance, comprising 12,000 video-annotation pairs, to provide a unified benchmark for training and evaluating systems to help visually-impaired individuals walk. Furthermore, a WalkVLM model is proposed, which employs chain of thought for hierarchical planning to generate concise but informative reminders and utilizes temporal-aware adaptive prediction to reduce the temporal redundancy of reminders. Finally, we have established a solid benchmark for blind walking task and verified the advantages of WalkVLM in stream video processing for this task compared to other VLMs. Our dataset and code are available at https://walkvlm2024.github.io.
♻ ☆ Rethinking High-speed Image Reconstruction Framework with Spike Camera AAAI2025
Spike cameras, as innovative neuromorphic devices, generate continuous spike streams to capture high-speed scenes with lower bandwidth and higher dynamic range than traditional RGB cameras. However, reconstructing high-quality images from the spike input under low-light conditions remains challenging. Conventional learning-based methods often rely on the synthetic dataset as the supervision for training. Still, these approaches falter when dealing with noisy spikes fired under the low-light environment, leading to further performance degradation in the real-world dataset. This phenomenon is primarily due to inadequate noise modelling and the domain gap between synthetic and real datasets, resulting in recovered images with unclear textures, excessive noise, and diminished brightness. To address these challenges, we introduce a novel spike-to-image reconstruction framework SpikeCLIP that goes beyond traditional training paradigms. Leveraging the CLIP model's powerful capability to align text and images, we incorporate the textual description of the captured scene and unpaired high-quality datasets as the supervision. Our experiments on real-world low-light datasets U-CALTECH and U-CIFAR demonstrate that SpikeCLIP significantly enhances texture details and the luminance balance of recovered images. Furthermore, the reconstructed images are well-aligned with the broader visual features needed for downstream tasks, ensuring more robust and versatile performance in challenging environments.
comment: Accepted by AAAI2025
♻ ☆ R2Det: Exploring Relaxed Rotation Equivariance in 2D object detection
Group Equivariant Convolution (GConv) empowers models to explore underlying symmetry in data, improving performance. However, real-world scenarios often deviate from ideal symmetric systems caused by physical permutation, characterized by non-trivial actions of a symmetry group, resulting in asymmetries that affect the outputs, a phenomenon known as Symmetry Breaking. Traditional GConv-based methods are constrained by rigid operational rules within group space, assuming data remains strictly symmetry after limited group transformations. This limitation makes it difficult to adapt to Symmetry-Breaking and non-rigid transformations. Motivated by this, we mainly focus on a common scenario: Rotational Symmetry-Breaking. By relaxing strict group transformations within Strict Rotation-Equivariant group $\mathbf{C}_n$, we redefine a Relaxed Rotation-Equivariant group $\mathbf{R}_n$ and introduce a novel Relaxed Rotation-Equivariant GConv (R2GConv) with only a minimal increase of $4n$ parameters compared to GConv. Based on R2GConv, we propose a Relaxed Rotation-Equivariant Network (R2Net) as the backbone and develop a Relaxed Rotation-Equivariant Object Detector (R2Det) for 2D object detection. Experimental results demonstrate the effectiveness of the proposed R2GConv in natural image classification, and R2Det achieves excellent performance in 2D object detection with improved generalization capabilities and robustness. The code is available in \texttt{https://github.com/wuer5/r2det}.
♻ ☆ PainDiffusion: Learning to Express Pain
Accurate pain expression synthesis is essential for improving clinical training and human-robot interaction. Current Robotic Patient Simulators (RPSs) lack realistic pain facial expressions, limiting their effectiveness in medical training. In this work, we introduce PainDiffusion, a generative model that synthesizes naturalistic facial pain expressions. Unlike traditional heuristic or autoregressive methods, PainDiffusion operates in a continuous latent space, ensuring smoother and more natural facial motion while supporting indefinite-length generation via diffusion forcing. Our approach incorporates intrinsic characteristics such as pain expressiveness and emotion, allowing for personalized and controllable pain expression synthesis. We train and evaluate our model using the BioVid HeatPain Database. Additionally, we integrate PainDiffusion into a robotic system to assess its applicability in real-time rehabilitation exercises. Qualitative studies with clinicians reveal that PainDiffusion produces realistic pain expressions, with a 31.2% (std 4.8%) preference rate against ground-truth recordings. Our results suggest that PainDiffusion can serve as a viable alternative to real patients in clinical training and simulation, bridging the gap between synthetic and naturalistic pain expression. Code and videos are available at: https://damtien444.github.io/paindf/
comment: 8 pages, 9 figures
♻ ☆ MedFuncta: Modality-Agnostic Representations Based on Efficient Neural Fields
Recent research in medical image analysis with deep learning almost exclusively focuses on grid- or voxel-based data representations. We challenge this common choice by introducing MedFuncta, a modality-agnostic continuous data representation based on neural fields. We demonstrate how to scale neural fields from single instances to large datasets by exploiting redundancy in medical signals and by applying an efficient meta-learning approach with a context reduction scheme. We further address the spectral bias in commonly used SIREN activations, by introducing an $\omega_0$-schedule, improving reconstruction quality and convergence speed. We validate our proposed approach on a large variety of medical signals of different dimensions and modalities (1D: ECG; 2D: Chest X-ray, Retinal OCT, Fundus Camera, Dermatoscope, Colon Histopathology, Cell Microscopy; 3D: Brain MRI, Lung CT) and successfully demonstrate that we can solve relevant downstream tasks on these representations. We additionally release a large-scale dataset of > 550k annotated neural fields to promote research in this direction.
comment: Project page: https://pfriedri.github.io/medfuncta-io/ Code and Dataset: https://github.com/pfriedri/medfuncta/
♻ ☆ Interpretable Interaction Modeling for Trajectory Prediction via Agent Selection and Physical Coefficient SP
A thorough understanding of the interaction between the target agent and surrounding agents is a prerequisite for accurate trajectory prediction. Although many methods have been explored, they assign correlation coefficients to surrounding agents in a purely learning-based manner. In this study, we present ASPILin, which manually selects interacting agents and replaces the attention scores in Transformer with a newly computed physical correlation coefficient, enhancing the interpretability of interaction modeling. Surprisingly, these simple modifications can significantly improve prediction performance and substantially reduce computational costs. We intentionally simplified our model in other aspects, such as map encoding. Remarkably, experiments conducted on the INTERACTION, highD, and CitySim datasets demonstrate that our method is efficient and straightforward, outperforming other state-of-the-art methods.
comment: code:https://github.com/kkk00714/ASPILin
♻ ☆ Augmenting Chest X-ray Datasets with Non-Expert Annotations
The advancement of machine learning algorithms in medical image analysis requires the expansion of training datasets. A popular and cost-effective approach is automated annotation extraction from free-text medical reports, primarily due to the high costs associated with expert clinicians annotating medical images, such as chest X-rays. However, it has been shown that the resulting datasets are susceptible to biases and shortcuts. Another strategy to increase the size of a dataset is crowdsourcing, a widely adopted practice in general computer vision with some success in medical image analysis. In a similar vein to crowdsourcing, we enhance two publicly available chest X-ray datasets by incorporating non-expert annotations. However, instead of using diagnostic labels, we annotate shortcuts in the form of tubes. We collect 3.5k chest drain annotations for NIH-CXR14, and 1k annotations for four different tube types in PadChest, and create the Non-Expert Annotations of Tubes in X-rays (NEATX) dataset. We train a chest drain detector with the non-expert annotations that generalizes well to expert labels. Moreover, we compare our annotations to those provided by experts and show "moderate" to "almost perfect" agreement. Finally, we present a pathology agreement study to raise awareness about the quality of ground truth annotations. We make our dataset available at https://zenodo.org/records/14944064 and our code available at https://github.com/purrlab/chestxr-label-reliability.
♻ ☆ Memory-Efficient Optical Flow via Radius-Distribution Orthogonal Cost Volume
The full 4D cost volume in Recurrent All-Pairs Field Transforms (RAFT) or global matching by Transformer achieves impressive performance for optical flow estimation. However, their memory consumption increases quadratically with input resolution, rendering them impractical for high-resolution images. In this paper, we present MeFlow, a novel memory-efficient method for high-resolution optical flow estimation. The key of MeFlow is a recurrent local orthogonal cost volume representation, which decomposes the 2D search space dynamically into two 1D orthogonal spaces, enabling our method to scale effectively to very high-resolution inputs. To preserve essential information in the orthogonal space, we utilize self attention to propagate feature information from the 2D space to the orthogonal space. We further propose a radius-distribution multi-scale lookup strategy to model the correspondences of large displacements at a negligible cost. We verify the efficiency and effectiveness of our method on the challenging Sintel and KITTI benchmarks, and real-world 4K ($2160\!\times\!3840$) images. Our method achieves competitive performance on both Sintel and KITTI benchmarks, while maintaining the highest memory efficiency on high-resolution inputs.
comment: 10 pages, 9 figures
♻ ☆ Deep Height Decoupling for Precise Vision-based 3D Occupancy Prediction
The task of vision-based 3D occupancy prediction aims to reconstruct 3D geometry and estimate its semantic classes from 2D color images, where the 2D-to-3D view transformation is an indispensable step. Most previous methods conduct forward projection, such as BEVPooling and VoxelPooling, both of which map the 2D image features into 3D grids. However, the current grid representing features within a certain height range usually introduces many confusing features that belong to other height ranges. To address this challenge, we present Deep Height Decoupling (DHD), a novel framework that incorporates explicit height prior to filter out the confusing features. Specifically, DHD first predicts height maps via explicit supervision. Based on the height distribution statistics, DHD designs Mask Guided Height Sampling (MGHS) to adaptively decouple the height map into multiple binary masks. MGHS projects the 2D image features into multiple subspaces, where each grid contains features within reasonable height ranges. Finally, a Synergistic Feature Aggregation (SFA) module is deployed to enhance the feature representation through channel and spatial affinities, enabling further occupancy refinement. On the popular Occ3D-nuScenes benchmark, our method achieves state-of-the-art performance even with minimal input frames. Source code is released at https://github.com/yanzq95/DHD.
♻ ☆ Dual Consolidation for Pre-Trained Model-Based Domain-Incremental Learning CVPR 2025
Domain-Incremental Learning (DIL) involves the progressive adaptation of a model to new concepts across different domains. While recent advances in pre-trained models provide a solid foundation for DIL, learning new concepts often results in the catastrophic forgetting of pre-trained knowledge. Specifically, sequential model updates can overwrite both the representation and the classifier with knowledge from the latest domain. Thus, it is crucial to develop a representation and corresponding classifier that accommodate all seen domains throughout the learning process. To this end, we propose DUal ConsolidaTion (Duct) to unify and consolidate historical knowledge at both the representation and classifier levels. By merging the backbone of different stages, we create a representation space suitable for multiple domains incrementally. The merged representation serves as a balanced intermediary that captures task-specific features from all seen domains. Additionally, to address the mismatch between consolidated embeddings and the classifier, we introduce an extra classifier consolidation process. Leveraging class-wise semantic information, we estimate the classifier weights of old domains within the latest embedding space. By merging historical and estimated classifiers, we align them with the consolidated embedding space, facilitating incremental classification. Extensive experimental results on four benchmark datasets demonstrate Duct's state-of-the-art performance. Code is available at https://github.com/Estrella-fugaz/CVPR25-Duct
comment: Accepted to CVPR 2025. Code is available at https://github.com/Estrella-fugaz/CVPR25-Duct
♻ ☆ DermaSynth: Rich Synthetic Image-Text Pairs Using Open Access Dermatology Datasets
A major barrier to developing vision large language models (LLMs) in dermatology is the lack of large image--text pairs dataset. We introduce DermaSynth, a dataset comprising of 92,020 synthetic image--text pairs curated from 45,205 images (13,568 clinical and 35,561 dermatoscopic) for dermatology-related clinical tasks. Leveraging state-of-the-art LLMs, using Gemini 2.0, we used clinically related prompts and self-instruct method to generate diverse and rich synthetic texts. Metadata of the datasets were incorporated into the input prompts by targeting to reduce potential hallucinations. The resulting dataset builds upon open access dermatological image repositories (DERM12345, BCN20000, PAD-UFES-20, SCIN, and HIBA) that have permissive CC-BY-4.0 licenses. We also fine-tuned a preliminary Llama-3.2-11B-Vision-Instruct model, DermatoLlama 1.0, on 5,000 samples. We anticipate this dataset to support and accelerate AI research in dermatology. Data and code underlying this work are accessible at https://github.com/abdurrahimyilmaz/DermaSynth.
comment: 12 pages, 4 figures
♻ ☆ Order Theory in the Context of Machine Learning NeurIPS
The paper ``Tropical Geometry of Deep Neural Networks'' by L. Zhang et al. introduces an equivalence between integer-valued neural networks (IVNN) with $\text{ReLU}_{t}$ and tropical rational functions, which come with a map to polytopes. Here, IVNN refers to a network with integer weights but real biases, and $\text{ReLU}_{t}$ is defined as $\text{ReLU}_{t}(x)=\max(x,t)$ for $t\in\mathbb{R}\cup\{-\infty\}$. For every poset with $n$ points, there exists a corresponding order polytope, i.e., a convex polytope in the unit cube $[0,1]^n$ whose coordinates obey the inequalities of the poset. We study neural networks whose associated polytope is an order polytope. We then explain how posets with four points induce neural networks that can be interpreted as $2\times 2$ convolutional filters. These poset filters can be added to any neural network, not only IVNN. Similarly to maxout, poset pooling filters update the weights of the neural network during backpropagation with more precision than average pooling, max pooling, or mixed pooling, without the need to train extra parameters. We report experiments that support our statements. We also define the structure of algebra over the operad of posets on poset neural networks and tropical polynomials. This formalism allows us to study the composition of poset neural network arquitectures and the effect on their corresponding Newton polytopes, via the introduction of the generalization of two operations on polytopes: the Minkowski sum and the convex envelope.
comment: We added experiments with ImageNet 100, and improved the exposition of the theory developed. Added examples. Poster presentation in NeurIPS WIML 2024, Talk in JMM 2025 section: Applied category theory II
♻ ☆ Smoothing Grounding and Reasoning for MLLM-Powered GUI Agents with Query-Oriented Pivot Tasks
Perception-enhanced pre-training, particularly through grounding techniques, is widely adopted to enhance the performance of graphical user interface (GUI) agents. However, in resource-constrained scenarios, the format discrepancy between coordinate-oriented grounding and action-oriented reasoning limits the effectiveness of grounding for reasoning tasks. To address this challenge, we propose a query-oriented pivot approach called query inference, which serves as a bridge between GUI grounding and reasoning. By inferring potential user queries from a screenshot and its associated element coordinates, query inference improves the understanding of coordinates while aligning more closely with reasoning tasks. Experimental results show that query inference outperforms previous grounding techniques under the same training data scale. Notably, query inference achieves comparable or even better performance to large-scale grounding-enhanced OS-Atlas with less than 0.1% of training data. Furthermore, we explore the impact of reasoning formats and demonstrate that integrating additional semantic information into the input further boosts reasoning performance. The code is publicly available at https://github.com/ZrW00/GUIPivot.
♻ ☆ A Multi-Sensor Fusion Approach for Rapid Orthoimage Generation in Large-Scale UAV Mapping
Rapid generation of large-scale orthoimages from Unmanned Aerial Vehicles (UAVs) has been a long-standing focus of research in the field of aerial mapping. A multi-sensor UAV system, integrating the Global Positioning System (GPS), Inertial Measurement Unit (IMU), 4D millimeter-wave radar and camera, can provide an effective solution to this problem. In this paper, we utilize multi-sensor data to overcome the limitations of conventional orthoimage generation methods in terms of temporal performance, system robustness, and geographic reference accuracy. A prior-pose-optimized feature matching method is introduced to enhance matching speed and accuracy, reducing the number of required features and providing precise references for the Structure from Motion (SfM) process. The proposed method exhibits robustness in low-texture scenes like farmlands, where feature matching is difficult. Experiments show that our approach achieves accurate feature matching orthoimage generation in a short time. The proposed drone system effectively aids in farmland detection and management.
♻ ☆ D$^2$-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models AAAI2025
Diffusion models have achieved cutting-edge performance in image generation. However, their lengthy denoising process and computationally intensive score estimation network impede their scalability in low-latency and resource-constrained scenarios. Post-training quantization (PTQ) compresses and accelerates diffusion models without retraining, but it inevitably introduces additional quantization noise, resulting in mean and variance deviations. In this work, we propose D2-DPM, a dual denoising mechanism aimed at precisely mitigating the adverse effects of quantization noise on the noise estimation network. Specifically, we first unravel the impact of quantization noise on the sampling equation into two components: the mean deviation and the variance deviation. The mean deviation alters the drift coefficient of the sampling equation, influencing the trajectory trend, while the variance deviation magnifies the diffusion coefficient, impacting the convergence of the sampling trajectory. The proposed D2-DPM is thus devised to denoise the quantization noise at each time step, and then denoise the noisy sample through the inverse diffusion iterations. Experimental results demonstrate that D2-DPM achieves superior generation quality, yielding a 1.42 lower FID than the full-precision model while achieving 3.99x compression and 11.67x bit-operation acceleration.
comment: 9 pages, 4 figures, acceptted by AAAI2025, the code is available at https://github.com/taylorjocelyn/d2-dpm
♻ ☆ InstaFace: Identity-Preserving Facial Editing with Single Image Inference
Facial appearance editing is crucial for digital avatars, AR/VR, and personalized content creation, driving realistic user experiences. However, preserving identity with generative models is challenging, especially in scenarios with limited data availability. Traditional methods often require multiple images and still struggle with unnatural face shifts, inconsistent hair alignment, or excessive smoothing effects. To overcome these challenges, we introduce a novel diffusion-based framework, InstaFace, to generate realistic images while preserving identity using only a single image. Central to InstaFace, we introduce an efficient guidance network that harnesses 3D perspectives by integrating multiple 3DMM-based conditionals without introducing additional trainable parameters. Moreover, to ensure maximum identity retention as well as preservation of background, hair, and other contextual features like accessories, we introduce a novel module that utilizes feature embeddings from a facial recognition model and a pre-trained vision-language model. Quantitative evaluations demonstrate that our method outperforms several state-of-the-art approaches in terms of identity preservation, photorealism, and effective control of pose, expression, and lighting.
♻ ☆ GenConViT: Deepfake Video Detection Using Generative Convolutional Vision Transformer
Deepfakes have raised significant concerns due to their potential to spread false information and compromise digital media integrity. Current deepfake detection models often struggle to generalize across a diverse range of deepfake generation techniques and video content. In this work, we propose a Generative Convolutional Vision Transformer (GenConViT) for deepfake video detection. Our model combines ConvNeXt and Swin Transformer models for feature extraction, and it utilizes Autoencoder and Variational Autoencoder to learn from the latent data distribution. By learning from the visual artifacts and latent data distribution, GenConViT achieves improved performance in detecting a wide range of deepfake videos. The model is trained and evaluated on DFDC, FF++, TM, DeepfakeTIMIT, and Celeb-DF (v$2$) datasets. The proposed GenConViT model demonstrates strong performance in deepfake video detection, achieving high accuracy across the tested datasets. While our model shows promising results in deepfake video detection by leveraging visual and latent features, we demonstrate that further work is needed to improve its generalizability, i.e., when encountering out-of-distribution data. Our model provides an effective solution for identifying a wide range of fake videos while preserving media integrity. The open-source code for GenConViT is available at https://github.com/erprogs/GenConViT.
comment: 11 pages, 4 figures
♻ ☆ AVD2: Accident Video Diffusion for Accident Video Description ICRA 2025
Traffic accidents present complex challenges for autonomous driving, often featuring unpredictable scenarios that hinder accurate system interpretation and responses. Nonetheless, prevailing methodologies fall short in elucidating the causes of accidents and proposing preventive measures due to the paucity of training data specific to accident scenarios. In this work, we introduce AVD2 (Accident Video Diffusion for Accident Video Description), a novel framework that enhances accident scene understanding by generating accident videos that aligned with detailed natural language descriptions and reasoning, resulting in the contributed EMM-AU (Enhanced Multi-Modal Accident Video Understanding) dataset. Empirical results reveal that the integration of the EMM-AU dataset establishes state-of-the-art performance across both automated metrics and human evaluations, markedly advancing the domains of accident analysis and prevention. Project resources are available at https://an-answer-tree.github.io
comment: ICRA 2025, Project Page: https://an-answer-tree.github.io/
♻ ☆ Replay Consolidation with Label Propagation for Continual Object Detection
Continual Learning (CL) aims to learn new data while remembering previously acquired knowledge. In contrast to CL for image classification, CL for Object Detection faces additional challenges such as the missing annotations problem. In this scenario, images from previous tasks may contain instances of unknown classes that could reappear as labeled in future tasks, leading to task interference in replay-based approaches. Consequently, most approaches in the literature have focused on distillation-based techniques, which are effective when there is a significant class overlap between tasks. In our work, we propose an alternative to distillation-based approaches with a novel approach called Replay Consolidation with Label Propagation for Object Detection (RCLPOD). RCLPOD enhances the replay memory by improving the quality of the stored samples through a technique that promotes class balance while also improving the quality of the ground truth associated with these samples through a technique called label propagation. RCLPOD outperforms existing techniques on well-established benchmarks such as VOC and COC. Moreover, our approach is developed to work with modern architectures like YOLOv8, making it suitable for dynamic, real-world applications such as autonomous driving and robotics, where continuous learning and resource efficiency are essential.
♻ ☆ Human Insights Driven Latent Space for Different Driving Perspectives: A Unified Encoder for Efficient Multi-Task Inference
Autonomous driving systems require a comprehensive understanding of the environment, achieved by extracting visual features essential for perception, planning, and control. However, models trained solely on single-task objectives or generic datasets often lack the contextual information needed for robust performance in complex driving scenarios. In this work, we propose a unified encoder trained on multiple computer vision tasks crucial for urban driving, including depth, pose, and 3D scene flow estimation, as well as semantic, instance, panoptic, and motion segmentation. By integrating these diverse visual cues-similar to human perceptual mechanisms-the encoder captures rich features that enhance navigation-related predictions. We evaluate the model on steering estimation as a downstream task, leveraging its dense latent space. To ensure efficient multi-task learning, we introduce a multi-scale feature network for pose estimation and apply knowledge distillation from a multi-backbone teacher model. Our findings highlight two key findings: (1) the unified encoder achieves competitive performance across all visual perception tasks, demonstrating strong generalization capabilities; and (2) for steering estimation, the frozen unified encoder-leveraging dense latent representations-outperforms both its fine-tuned counterpart and the same frozen model pretrained on generic datasets like ImageNet. These results underline the significance of task-specific visual features and demonstrate the promise of multi-task learning in advancing autonomous driving systems. More details and the pretrained model are available at https://hi-computervision.github.io/uni-encoder/.
♻ ☆ Assessing Robustness via Score-Based Adversarial Image Generation
Most adversarial attacks and defenses focus on perturbations within small $\ell_p$-norm constraints. However, $\ell_p$ threat models cannot capture all relevant semantics-preserving perturbations, and hence, the scope of robustness evaluations is limited. In this work, we introduce Score-Based Adversarial Generation (ScoreAG), a novel framework that leverages the advancements in score-based generative models to generate unrestricted adversarial examples that overcome the limitations of $\ell_p$-norm constraints. Unlike traditional methods, ScoreAG maintains the core semantics of images while generating adversarial examples, either by transforming existing images or synthesizing new ones entirely from scratch. We further exploit the generative capability of ScoreAG to purify images, empirically enhancing the robustness of classifiers. Our extensive empirical evaluation demonstrates that ScoreAG improves upon the majority of state-of-the-art attacks and defenses across multiple benchmarks. This work highlights the importance of investigating adversarial examples bounded by semantics rather than $\ell_p$-norm constraints. ScoreAG represents an important step towards more encompassing robustness assessments.
♻ ☆ Unposed Sparse Views Room Layout Reconstruction in the Age of Pretrain Model ICLR 2025
Room layout estimation from multiple-perspective images is poorly investigated due to the complexities that emerge from multi-view geometry, which requires muti-step solutions such as camera intrinsic and extrinsic estimation, image matching, and triangulation. However, in 3D reconstruction, the advancement of recent 3D foundation models such as DUSt3R has shifted the paradigm from the traditional multi-step structure-from-motion process to an end-to-end single-step approach. To this end, we introduce Plane-DUSt3R, a novel method for multi-view room layout estimation leveraging the 3D foundation model DUSt3R. Plane-DUSt3R incorporates the DUSt3R framework and fine-tunes on a room layout dataset (Structure3D) with a modified objective to estimate structural planes. By generating uniform and parsimonious results, Plane-DUSt3R enables room layout estimation with only a single post-processing step and 2D detection results. Unlike previous methods that rely on single-perspective or panorama image, Plane-DUSt3R extends the setting to handle multiple-perspective images. Moreover, it offers a streamlined, end-to-end solution that simplifies the process and reduces error accumulation. Experimental results demonstrate that Plane-DUSt3R not only outperforms state-of-the-art methods on the synthetic dataset but also proves robust and effective on in the wild data with different image styles such as cartoon. Our code is available at: https://github.com/justacar/Plane-DUSt3R
comment: Accepted by ICLR 2025. Github page:https://github.com/justacar/Plane-DUSt3R
♻ ☆ A Simple and Effective Reinforcement Learning Method for Text-to-Image Diffusion Fine-tuning
Reinforcement learning ( RL)-based fine-tuning has emerged as a powerful approach for aligning diffusion models with black-box objectives. Proximal policy optimization (PPO) is the most popular choice of method for policy optimization. While effective in terms of performance, PPO is highly sensitive to hyper-parameters and involves substantial computational overhead. REINFORCE, on the other hand, mitigates some computational complexities such as high memory overhead and sensitive hyper-parameter tuning, but has suboptimal performance due to high-variance and sample inefficiency. While the variance of the REINFORCE can be reduced by sampling multiple actions per input prompt and using a baseline correction term, it still suffers from sample inefficiency. To address these challenges, we systematically analyze the efficiency-effectiveness trade-off between REINFORCE and PPO, and propose leave-one-out PPO ( LOOP), a novel RL for diffusion fine-tuning method. LOOP combines variance reduction techniques from REINFORCE, such as sampling multiple actions per input prompt and a baseline correction term, with the robustness and sample efficiency of PPO via clipping and importance sampling. Our results demonstrate that LOOP effectively improves diffusion models on various black-box objectives, and achieves a better balance between computational efficiency and performance.
♻ ☆ OpenFly: A Versatile Toolchain and Large-scale Benchmark for Aerial Vision-Language Navigation
Vision-Language Navigation (VLN) aims to guide agents through an environment by leveraging both language instructions and visual cues, playing a pivotal role in embodied AI. Indoor VLN has been extensively studied, whereas outdoor aerial VLN remains underexplored. The potential reason is that outdoor aerial view encompasses vast areas, making data collection more challenging, which results in a lack of benchmarks. To address this problem, we propose OpenFly, a platform comprising a versatile toolchain and large-scale benchmark for aerial VLN. Firstly, we develop a highly automated toolchain for data collection, enabling automatic point cloud acquisition, scene semantic segmentation, flight trajectory creation, and instruction generation. Secondly, based on the toolchain, we construct a large-scale aerial VLN dataset with 100k trajectories, covering diverse heights and lengths across 18 scenes. The corresponding visual data are generated using various rendering engines and advanced techniques, including Unreal Engine, GTA V, Google Earth, and 3D Gaussian Splatting (3D GS). All data exhibit high visual quality. Particularly, 3D GS supports real-to-sim rendering, further enhancing the realism of the dataset. Thirdly, we propose OpenFly-Agent, a keyframe-aware VLN model, which takes language instructions, current observations, and historical keyframes as input, and outputs flight actions directly. Extensive analyses and experiments are conducted, showcasing the superiority of our OpenFly platform and OpenFly-Agent. The toolchain, dataset, and codes will be open-sourced.
♻ ☆ Non-rigid Structure-from-Motion: Temporally-smooth Procrustean Alignment and Spatially-variant Deformation Modeling CVPR 2024
Even though Non-rigid Structure-from-Motion (NRSfM) has been extensively studied and great progress has been made, there are still key challenges that hinder their broad real-world applications: 1) the inherent motion/rotation ambiguity requires either explicit camera motion recovery with extra constraint or complex Procrustean Alignment; 2) existing low-rank modeling of the global shape can over-penalize drastic deformations in the 3D shape sequence. This paper proposes to resolve the above issues from a spatial-temporal modeling perspective. First, we propose a novel Temporally-smooth Procrustean Alignment module that estimates 3D deforming shapes and adjusts the camera motion by aligning the 3D shape sequence consecutively. Our new alignment module remedies the requirement of complex reference 3D shape during alignment, which is more conductive to non-isotropic deformation modeling. Second, we propose a spatial-weighted approach to enforce the low-rank constraint adaptively at different locations to accommodate drastic spatially-variant deformation reconstruction better. Our modeling outperform existing low-rank based methods, and extensive experiments across different datasets validate the effectiveness of our method.
comment: Accepted by CVPR 2024; The new version adds additional experiments and corrects typos
♻ ☆ TwinLiteNetPlus: A Stronger Model for Real-time Drivable Area and Lane Segmentation
Semantic segmentation is crucial for autonomous driving, particularly for Drivable Area and Lane Segmentation, ensuring safety and navigation. To address the high computational costs of current state-of-the-art (SOTA) models, this paper introduces TwinLiteNetPlus (TwinLiteNet$^+$), a model adept at balancing efficiency and accuracy. TwinLiteNet$^+$ incorporates standard and depth-wise separable dilated convolutions, reducing complexity while maintaining high accuracy. It is available in four configurations, from the robust 1.94 million-parameter TwinLiteNet$^+_{\text{Large}}$ to the ultra-compact 34K-parameter TwinLiteNet$^+_{\text{Nano}}$. Notably, TwinLiteNet$^+_{\text{Large}}$ attains a 92.9\% mIoU for Drivable Area Segmentation and a 34.2\% IoU for Lane Segmentation. These results notably outperform those of current SOTA models while requiring a computational cost that is approximately 11 times lower in terms of Floating Point Operations (FLOPs) compared to the existing SOTA model. Extensively tested on various embedded devices, TwinLiteNet$^+$ demonstrates promising latency and power efficiency, underscoring its suitability for real-world autonomous vehicle applications.
♻ ☆ Unified Video Action Model
A unified video and action model holds significant promise for robotics, where videos provide rich scene information for action prediction, and actions provide dynamics information for video prediction. However, effectively combining video generation and action prediction remains challenging, and current video generation-based methods struggle to match the performance of direct policy learning in action accuracy and inference speed. To bridge this gap, we introduce the Unified Video Action model (UVA), which jointly optimizes video and action predictions to achieve both high accuracy and efficient action inference. The key lies in learning a joint video-action latent representation and decoupling video-action decoding. The joint latent representation bridges the visual and action domains, effectively modeling the relationship between video and action sequences. Meanwhile, the decoupled decoding, powered by two lightweight diffusion heads, enables high-speed action inference by bypassing video generation during inference. Such a unified framework further enables versatile functionality through masked input training. By selectively masking actions or videos, a single model can tackle diverse tasks beyond policy learning, such as forward and inverse dynamics modeling and video generation. Via an extensive set of experiments, we demonstrate that UVA can serve as a general-purpose solution for a wide range of robotics tasks, such as policy learning, forward/inverse dynamics and video observation prediction, without compromising performance compared to methods tailored for specific applications. Results are best viewed on https://unified-video-action-model.github.io/.
comment: Project website: https://unified-video-action-model.github.io/
♻ ☆ SeqAfford: Sequential 3D Affordance Reasoning via Multimodal Large Language Model
3D affordance segmentation aims to link human instructions to touchable regions of 3D objects for embodied manipulations. Existing efforts typically adhere to single-object, single-affordance paradigms, where each affordance type or explicit instruction strictly corresponds to a specific affordance region and are unable to handle long-horizon tasks. Such a paradigm cannot actively reason about complex user intentions that often imply sequential affordances. In this paper, we introduce the Sequential 3D Affordance Reasoning task, which extends the traditional paradigm by reasoning from cumbersome user intentions and then decomposing them into a series of segmentation maps. Toward this, we construct the first instruction-based affordance segmentation benchmark that includes reasoning over both single and sequential affordances, comprising 180K instruction-point cloud pairs. Based on the benchmark, we propose our model, SeqAfford, to unlock the 3D multi-modal large language model with additional affordance segmentation abilities, which ensures reasoning with world knowledge and fine-grained affordance grounding in a cohesive framework. We further introduce a multi-granular language-point integration module to endow 3D dense prediction. Extensive experimental evaluations show that our model excels over well-established methods and exhibits open-world generalization with sequential reasoning abilities.
♻ ☆ VideoRAG: Retrieval-Augmented Generation over Video Corpus
Retrieval-Augmented Generation (RAG) is a powerful strategy for improving the factual accuracy of models by retrieving external knowledge relevant to queries and incorporating it into the generation process. However, existing approaches primarily focus on text, with some recent advancements considering images, and they largely overlook videos, a rich source of multimodal knowledge capable of representing contextual details more effectively than any other modality. While very recent studies explore the use of videos in response generation, they either predefine query-associated videos without retrieval or convert videos into textual descriptions losing multimodal richness. To tackle these, we introduce VideoRAG, a framework that not only dynamically retrieves videos based on their relevance with queries but also utilizes both visual and textual information. The operation of VideoRAG is powered by recent Large Video Language Models (LVLMs), which enable the direct processing of video content to represent it for retrieval and the seamless integration of retrieved videos jointly with queries for response generation. Also, inspired by that the context size of LVLMs may not be sufficient to process all frames in extremely long videos and not all frames are equally important, we introduce a video frame selection mechanism to extract the most informative subset of frames, along with a strategy to extract textual information from videos (as it can aid the understanding of video content) when their subtitles are not available. We experimentally validate the effectiveness of VideoRAG, showcasing that it is superior to relevant baselines. Code is available at https://github.com/starsuzi/VideoRAG.
♻ ☆ Generalized Gaussian Model for Learned Image Compression
In learned image compression, probabilistic models play an essential role in characterizing the distribution of latent variables. The Gaussian model with mean and scale parameters has been widely used for its simplicity and effectiveness. Probabilistic models with more parameters, such as the Gaussian mixture models, can fit the distribution of latent variables more precisely, but the corresponding complexity is higher. To balance the compression performance and complexity, we extend the Gaussian model to the generalized Gaussian family for more flexible latent distribution modeling, introducing only one additional shape parameter beta than the Gaussian model. To enhance the performance of the generalized Gaussian model by alleviating the train-test mismatch, we propose improved training methods, including beta-dependent lower bounds for scale parameters and gradient rectification. Our proposed generalized Gaussian model, coupled with the improved training methods, is demonstrated to outperform the Gaussian and Gaussian mixture models on a variety of learned image compression networks.
comment: 19 pages, 16 figures
♻ ☆ SelaFD:Seamless Adaptation of Vision Transformer Fine-tuning for Radar-based Human Activity Recognition
Human Activity Recognition (HAR) such as fall detection has become increasingly critical due to the aging population, necessitating effective monitoring systems to prevent serious injuries and fatalities associated with falls. This study focuses on fine-tuning the Vision Transformer (ViT) model specifically for HAR using radar-based Time-Doppler signatures. Unlike traditional image datasets, these signals present unique challenges due to their non-visual nature and the high degree of similarity among various activities. Directly fine-tuning the ViT with all parameters proves suboptimal for this application. To address this challenge, we propose a novel approach that employs Low-Rank Adaptation (LoRA) fine-tuning in the weight space to facilitate knowledge transfer from pre-trained ViT models. Additionally, to extract fine-grained features, we enhance feature representation through the integration of a serial-parallel adapter in the feature space. Our innovative joint fine-tuning method, tailored for radar-based Time-Doppler signatures, significantly improves HAR accuracy, surpassing existing state-of-the-art methodologies in this domain. Our code is released at https://github.com/wangyijunlyy/SelaFD.
♻ ☆ PolaFormer: Polarity-aware Linear Attention for Vision Transformers
Linear attention has emerged as a promising alternative to softmax-based attention, leveraging kernelized feature maps to reduce complexity from quadratic to linear in sequence length. However, the non-negative constraint on feature maps and the relaxed exponential function used in approximation lead to significant information loss compared to the original query-key dot products, resulting in less discriminative attention maps with higher entropy. To address the missing interactions driven by negative values in query-key pairs, we propose a polarity-aware linear attention mechanism that explicitly models both same-signed and opposite-signed query-key interactions, ensuring comprehensive coverage of relational information. Furthermore, to restore the spiky properties of attention maps, we provide a theoretical analysis proving the existence of a class of element-wise functions (with positive first and second derivatives) that can reduce entropy in the attention distribution. For simplicity, and recognizing the distinct contributions of each dimension, we employ a learnable power function for rescaling, allowing strong and weak attention signals to be effectively separated. Extensive experiments demonstrate that the proposed PolaFormer improves performance on various vision tasks, enhancing both expressiveness and efficiency by up to 4.6%.
♻ ☆ Evaluating the Robustness of LiDAR Point Cloud Tracking Against Adversarial Attack
In this study, we delve into the robustness of neural network-based LiDAR point cloud tracking models under adversarial attacks, a critical aspect often overlooked in favor of performance enhancement. These models, despite incorporating advanced architectures like Transformer or Bird's Eye View (BEV), tend to neglect robustness in the face of challenges such as adversarial attacks, domain shifts, or data corruption. We instead focus on the robustness of the tracking models under the threat of adversarial attacks. We begin by establishing a unified framework for conducting adversarial attacks within the context of 3D object tracking, which allows us to thoroughly investigate both white-box and black-box attack strategies. For white-box attacks, we tailor specific loss functions to accommodate various tracking paradigms and extend existing methods such as FGSM, C\&W, and PGD to the point cloud domain. In addressing black-box attack scenarios, we introduce a novel transfer-based approach, the Target-aware Perturbation Generation (TAPG) algorithm, with the dual objectives of achieving high attack performance and maintaining low perceptibility. This method employs a heuristic strategy to enforce sparse attack constraints and utilizes random sub-vector factorization to bolster transferability. Our experimental findings reveal a significant vulnerability in advanced tracking methods when subjected to both black-box and white-box attacks, underscoring the necessity for incorporating robustness against adversarial attacks into the design of LiDAR point cloud tracking models. Notably, compared to existing methods, the TAPG also strikes an optimal balance between the effectiveness of the attack and the concealment of the perturbations.
♻ ☆ Tera-MIND: Tera-scale mouse brain simulation via spatial mRNA-guided diffusion
Holistic 3D modeling of molecularly defined brain structures is crucial for understanding complex brain functions. Emerging tissue profiling technologies enable the construction of a comprehensive atlas of the mammalian brain with sub-cellular resolution and spatially resolved gene expression data. However, such tera-scale volumetric datasets present significant computational challenges in understanding complex brain functions within their native 3D spatial context. Here, we propose the novel generative approach $\textbf{Tera-MIND}$, which can simulate $\textbf{Tera}$-scale $\textbf{M}$ouse bra$\textbf{IN}$s in 3D using a patch-based and boundary-aware $\textbf{D}$iffusion model. Taking spatial transcriptomic data as the conditional input, we generate virtual mouse brains with comprehensive cellular morphological detail at teravoxel scale. Through the lens of 3D $gene$-$gene$ self-attention, we identify spatial molecular interactions for key transcriptomic pathways in the murine brain, exemplified by glutamatergic and dopaminergic neuronal systems. Importantly, these $in$-$silico$ biological findings are consistent and reproducible across three tera-scale virtual mouse brains. Therefore, Tera-MIND showcases a promising path toward efficient and generative simulations of whole organ systems for biomedical research. Project website: https://musikisomorphie.github.io/Tera-MIND.html
♻ ☆ An Efficient 3D Convolutional Neural Network with Channel-wise, Spatial-grouped, and Temporal Convolutions
There has been huge progress on video action recognition in recent years. However, many works focus on tweaking existing 2D backbones due to the reliance of ImageNet pretraining, which restrains the models from achieving higher efficiency for video recognition. In this work we introduce a simple and very efficient 3D convolutional neural network for video action recognition. The design of the building block consists of a channel-wise convolution, followed by a spatial group convolution, and finally a temporal convolution. We evaluate the performance and efficiency of our proposed network on several video action recognition datasets by directly training on the target dataset without relying on pertaining. On Something-Something-V1&V2, Kinetics-400 and Multi-Moments in Time, our network can match or even surpass the performance of other models which are several times larger. On the fine-grained action recognition dataset FineGym, we beat the previous state-of-the-art accuracy achieved with 2-stream methods by more than 5% using only RGB input.
♻ ☆ Residual Kolmogorov-Arnold Network for Enhanced Deep Learning
Despite their immense success, deep neural networks (CNNs) are costly to train, while modern architectures can retain hundreds of convolutional layers in network depth. Standard convolutional operations are fundamentally limited by their linear nature along with fixed activations, where multiple layers are needed to learn complex patterns, making this approach computationally inefficient and prone to optimization difficulties. As a result, we introduce RKAN (Residual Kolmogorov-Arnold Network), which could be easily implemented into stages of traditional networks, such as ResNet. The module also integrates polynomial feature transformation that provides the expressive power of many convolutional layers through learnable, non-linear feature refinement. Our proposed RKAN module offers consistent improvements over the base models on various well-known benchmark datasets, such as CIFAR-100, Food-101, and ImageNet.
comment: Code is available at https://github.com/withray/residualKAN.git
♻ ☆ Frame-Voyager: Learning to Query Frames for Video Large Language Models ICLR 2025
Video Large Language Models (Video-LLMs) have made remarkable progress in video understanding tasks. However, they are constrained by the maximum length of input tokens, making it impractical to input entire videos. Existing frame selection approaches, such as uniform frame sampling and text-frame retrieval, fail to account for the information density variations in the videos or the complex instructions in the tasks, leading to sub-optimal performance. In this paper, we propose Frame-Voyager that learns to query informative frame combinations, based on the given textual queries in the task. To train Frame-Voyager, we introduce a new data collection and labeling pipeline, by ranking frame combinations using a pre-trained Video-LLM. Given a video of M frames, we traverse its T-frame combinations, feed them into a Video-LLM, and rank them based on Video-LLM's prediction losses. Using this ranking as supervision, we train Frame-Voyager to query the frame combinations with lower losses. In experiments, we evaluate Frame-Voyager on four Video Question Answering benchmarks by plugging it into two different Video-LLMs. The experimental results demonstrate that Frame-Voyager achieves impressive results in all settings, highlighting its potential as a plug-and-play solution for Video-LLMs.
comment: ICLR 2025, Camera-ready Version
♻ ☆ MS-Diffusion: Multi-subject Zero-shot Image Personalization with Layout Guidance
Recent advancements in text-to-image generation models have dramatically enhanced the generation of photorealistic images from textual prompts, leading to an increased interest in personalized text-to-image applications, particularly in multi-subject scenarios. However, these advances are hindered by two main challenges: firstly, the need to accurately maintain the details of each referenced subject in accordance with the textual descriptions; and secondly, the difficulty in achieving a cohesive representation of multiple subjects in a single image without introducing inconsistencies. To address these concerns, our research introduces the MS-Diffusion framework for layout-guided zero-shot image personalization with multi-subjects. This innovative approach integrates grounding tokens with the feature resampler to maintain detail fidelity among subjects. With the layout guidance, MS-Diffusion further improves the cross-attention to adapt to the multi-subject inputs, ensuring that each subject condition acts on specific areas. The proposed multi-subject cross-attention orchestrates harmonious inter-subject compositions while preserving the control of texts. Comprehensive quantitative and qualitative experiments affirm that this method surpasses existing models in both image and text fidelity, promoting the development of personalized text-to-image generation. The project page is https://MS-Diffusion.github.io.
♻ ☆ Quantized Spike-driven Transformer ICLR 2025
Spiking neural networks are emerging as a promising energy-efficient alternative to traditional artificial neural networks due to their spike-driven paradigm. However, recent research in the SNN domain has mainly focused on enhancing accuracy by designing large-scale Transformer structures, which typically rely on substantial computational resources, limiting their deployment on resource-constrained devices. To overcome this challenge, we propose a quantized spike-driven Transformer baseline (QSD-Transformer), which achieves reduced resource demands by utilizing a low bit-width parameter. Regrettably, the QSD-Transformer often suffers from severe performance degradation. In this paper, we first conduct empirical analysis and find that the bimodal distribution of quantized spike-driven self-attention (Q-SDSA) leads to spike information distortion (SID) during quantization, causing significant performance degradation. To mitigate this issue, we take inspiration from mutual information entropy and propose a bi-level optimization strategy to rectify the information distribution in Q-SDSA. Specifically, at the lower level, we introduce an information-enhanced LIF to rectify the information distribution in Q-SDSA. At the upper level, we propose a fine-grained distillation scheme for the QSD-Transformer to align the distribution in Q-SDSA with that in the counterpart ANN. By integrating the bi-level optimization strategy, the QSD-Transformer can attain enhanced energy efficiency without sacrificing its high-performance advantage. For instance, when compared to the prior SNN benchmark on ImageNet, the QSD-Transformer achieves 80.3% top-1 accuracy, accompanied by significant reductions of 6.0$\times$ and 8.1$\times$ in power consumption and model size, respectively. Code is available at https://github.com/bollossom/QSD-Transformer.
comment: Accepted by ICLR 2025
♻ ☆ Abstract Rendering: Computing All that is Seen in Gaussian Splat Scenes
We introduce abstract rendering, a method for computing a set of images by rendering a scene from a continuously varying range of camera positions. The resulting abstract image-which encodes an infinite collection of possible renderings-is represented using constraints on the image matrix, enabling rigorous uncertainty propagation through the rendering process. This capability is particularly valuable for the formal verification of vision-based autonomous systems and other safety-critical applications. Our approach operates on Gaussian splat scenes, an emerging representation in computer vision and robotics. We leverage efficient piecewise linear bound propagation to abstract fundamental rendering operations, while addressing key challenges that arise in matrix inversion and depth sorting-two operations not directly amenable to standard approximations. To handle these, we develop novel linear relational abstractions that maintain precision while ensuring computational efficiency. These abstractions not only power our abstract rendering algorithm but also provide broadly applicable tools for other rendering problems. Our implementation, AbstractSplat, is optimized for scalability, handling up to 750k Gaussians while allowing users to balance memory and runtime through tile and batch-based computation. Compared to the only existing abstract image method for mesh-based scenes, AbstractSplat achieves 2-14x speedups while preserving precision. Our results demonstrate that continuous camera motion, rotations, and scene variations can be rigorously analyzed at scale, making abstract rendering a powerful tool for uncertainty-aware vision applications.
♻ ☆ Floorplan-SLAM: A Real-Time, High-Accuracy, and Long-Term Multi-Session Point-Plane SLAM for Efficient Floorplan Reconstruction
Floorplan reconstruction provides structural priors essential for reliable indoor robot navigation and high-level scene understanding. However, existing approaches either require time-consuming offline processing with a complete map, or rely on expensive sensors and substantial computational resources. To address the problems, we propose Floorplan-SLAM, which incorporates floorplan reconstruction tightly into a multi-session SLAM system by seamlessly interacting with plane extraction, pose estimation, and back-end optimization, achieving real-time, high-accuracy, and long-term floorplan reconstruction using only a stereo camera. Specifically, we present a robust plane extraction algorithm that operates in a compact plane parameter space and leverages spatially complementary features to accurately detect planar structures, even in weakly textured scenes. Furthermore, we propose a floorplan reconstruction module tightly coupled with the SLAM system, which uses continuously optimized plane landmarks and poses to formulate and solve a novel optimization problem, thereby enabling real-time incremental floorplan reconstruction. Note that by leveraging the map merging capability of multi-session SLAM, our method supports long-term floorplan reconstruction across multiple sessions without redundant data collection. Experiments on the VECtor and the self-collected datasets indicate that Floorplan-SLAM significantly outperforms state-of-the-art methods in terms of plane extraction robustness, pose estimation accuracy, and floorplan reconstruction fidelity and speed, achieving real-time performance at 25-45 FPS without GPU acceleration, which reduces the floorplan reconstruction time for a 1000 square meters scene from over 10 hours to just 9.44 minutes.
♻ ☆ NoPain: No-box Point Cloud Attack via Optimal Transport Singular Boundary
Adversarial attacks exploit the vulnerability of deep models against adversarial samples. Existing point cloud attackers are tailored to specific models, iteratively optimizing perturbations based on gradients in either a white-box or black-box setting. Despite their promising attack performance, they often struggle to produce transferable adversarial samples due to overfitting the specific parameters of surrogate models. To overcome this issue, we shift our focus to the data distribution itself and introduce a novel approach named NoPain, which employs optimal transport (OT) to identify the inherent singular boundaries of the data manifold for cross-network point cloud attacks. Specifically, we first calculate the OT mapping from noise to the target feature space, then identify singular boundaries by locating non-differentiable positions. Finally, we sample along singular boundaries to generate adversarial point clouds. Once the singular boundaries are determined, NoPain can efficiently produce adversarial samples without the need of iterative updates or guidance from the surrogate classifiers. Extensive experiments demonstrate that the proposed end-to-end method outperforms baseline approaches in terms of both transferability and efficiency, while also maintaining notable advantages even against defense strategies. The source code will be publicly available.
♻ ☆ PatternPaint: Practical Layout Pattern Generation Using Diffusion-Based Inpainting
Generating diverse VLSI layout patterns is essential for various downstream tasks in design for manufacturing, as design rules continually evolve during the development of new technology nodes. However, existing training-based methods for layout pattern generation rely on large datasets. In practical scenarios, especially when developing a new technology node, obtaining such extensive layout data is challenging. Consequently, training models with large datasets becomes impractical, limiting the scalability and adaptability of prior approaches. To this end, we propose PatternPaint, a diffusion-based framework capable of generating legal patterns with limited design-rule-compliant training samples. PatternPaint simplifies complex layout pattern generation into a series of inpainting processes with a template-based denoising scheme. Furthermore, we perform few-shot finetuning on a pretrained image foundation model with only 20 design-rule-compliant samples. Experimental results show that using a sub-3nm technology node (Intel 18A), our model is the only one that can generate legal patterns in complex 2D metal interconnect design rule settings among all previous works and achieves a high diversity score. Additionally, our few-shot finetuning can boost the legality rate with 1.87X improvement compared to the original pretrained model. As a result, we demonstrate a production-ready approach for layout pattern generation in developing new technology nodes.
♻ ☆ VIRES: Video Instance Repainting via Sketch and Text Guided Generation
We introduce VIRES, a video instance repainting method with sketch and text guidance, enabling video instance repainting, replacement, generation, and removal. Existing approaches struggle with temporal consistency and accurate alignment with the provided sketch sequence. VIRES leverages the generative priors of text-to-video models to maintain temporal consistency and produce visually pleasing results. We propose the Sequential ControlNet with the standardized self-scaling, which effectively extracts structure layouts and adaptively captures high-contrast sketch details. We further augment the diffusion transformer backbone with the sketch attention to interpret and inject fine-grained sketch semantics. A sketch-aware encoder ensures that repainted results are aligned with the provided sketch sequence. Additionally, we contribute the VireSet, a dataset with detailed annotations tailored for training and evaluating video instance editing methods. Experimental results demonstrate the effectiveness of VIRES, which outperforms state-of-the-art methods in visual quality, temporal consistency, condition alignment, and human ratings. Project page:https://suimuc.github.io/suimu.github.io/projects/VIRES/
♻ ☆ KiVA: Kid-inspired Visual Analogies for Testing Large Multimodal Models
This paper investigates visual analogical reasoning in large multimodal models (LMMs) compared to human adults and children. A "visual analogy" is an abstract rule inferred from one image and applied to another. While benchmarks exist for testing visual reasoning in LMMs, they require advanced skills and omit basic visual analogies that even young children can make. Inspired by developmental psychology, we propose a new benchmark of 4,300 visual transformations of everyday objects to test LMMs on visual analogical reasoning and compare them to children (ages three to five) and to adults. We structure the evaluation into three stages: identifying what changed (e.g., color, number, etc.), how it changed (e.g., added one object), and applying the rule to new scenarios. Our findings show that while GPT-o1, GPT-4V, LLaVA-1.5, and MANTIS identify the "what" effectively, they struggle with quantifying the "how" and extrapolating this rule to new objects. In contrast, children and adults exhibit much stronger analogical reasoning at all three stages. Additionally, the strongest tested model, GPT-o1, performs better in tasks involving simple surface-level visual attributes like color and size, correlating with quicker human adult response times. Conversely, more complex tasks such as number, rotation, and reflection, which necessitate extensive cognitive processing and understanding of extrinsic spatial properties in the physical world, present more significant challenges. Altogether, these findings highlight the limitations of training models on data that primarily consists of 2D images and text.
comment: 10 pages. For the KiVA benchmark, see https://github.com/ey242/KiVA
♻ ☆ SAR-W-MixMAE: SAR Foundation Model Training Using Backscatter Power Weighting
Foundation model approaches such as masked auto-encoders (MAE) or its variations are now being successfully applied to satellite imagery. Most of the ongoing technical validation of foundation models have been applied to optical images like RGB or multi-spectral images. Due to difficulty in semantic labeling to create datasets and higher noise content with respect to optical images, Synthetic Aperture Radar (SAR) data has not been explored a lot in the field for foundation models. Therefore, in this work as a pre-training approach, we explored masked auto-encoder, specifically MixMAE on Sentinel-1 SAR images and its impact on SAR image classification tasks. Moreover, we proposed to use the physical characteristic of SAR data for applying weighting parameter on the auto-encoder training loss (MSE) to reduce the effect of speckle noise and very high values on the SAR images. Proposed SAR intensity-based weighting of the reconstruction loss demonstrates promising results both on SAR pre-training and downstream tasks specifically on flood detection compared with the baseline model.
comment: 5 pages, 1 figure
♻ ☆ Accelerating Auto-regressive Text-to-Image Generation with Training-free Speculative Jacobi Decoding ICLR 2025
The current large auto-regressive models can generate high-quality, high-resolution images, but these models require hundreds or even thousands of steps of next-token prediction during inference, resulting in substantial time consumption. In existing studies, Jacobi decoding, an iterative parallel decoding algorithm, has been used to accelerate the auto-regressive generation and can be executed without training. However, the Jacobi decoding relies on a deterministic criterion to determine the convergence of iterations. Thus, it works for greedy decoding but is incompatible with sampling-based decoding which is crucial for visual quality and diversity in the current auto-regressive text-to-image generation. In this paper, we propose a training-free probabilistic parallel decoding algorithm, Speculative Jacobi Decoding (SJD), to accelerate auto-regressive text-to-image generation. By introducing a probabilistic convergence criterion, our SJD accelerates the inference of auto-regressive text-to-image generation while maintaining the randomness in sampling-based token decoding and allowing the model to generate diverse images. Specifically, SJD facilitates the model to predict multiple tokens at each step and accepts tokens based on the probabilistic criterion, enabling the model to generate images with fewer steps than the conventional next-token-prediction paradigm. We also investigate the token initialization strategies that leverage the spatial locality of visual data to further improve the acceleration ratio under specific scenarios. We conduct experiments for our proposed SJD on multiple auto-regressive text-to-image generation models, showing the effectiveness of model acceleration without sacrificing the visual quality. The code of our work is available here: https://github.com/tyshiwo1/Accelerating-T2I-AR-with-SJD/.
comment: ICLR 2025; Codes: https://github.com/tyshiwo1/Accelerating-T2I-AR-with-SJD/
♻ ☆ Fence Theorem: Towards Dual-Objective Semantic-Structure Isolation in Preprocessing Phase for 3D Anomaly Detection
3D anomaly detection (AD) is prominent but difficult due to lacking a unified theoretical foundation for preprocessing design. We establish the Fence Theorem, formalizing preprocessing as a dual-objective semantic isolator: (1) mitigating cross-semantic interference to the greatest extent feasible and (2) confining anomaly judgments to aligned semantic spaces wherever viable, thereby establishing intra-semantic comparability. Any preprocessing approach achieves this goal through a two-stage process of Emantic-Division and Spatial-Constraints stage. Through systematic deconstruction, we theoretically and experimentally subsume existing preprocessing methods under this theorem via tripartite evidence: qualitative analyses, quantitative studies, and mathematical proofs. Guided by the Fence Theorem, we implement Patch3D, consisting of Patch-Cutting and Patch-Matching modules, to segment semantic spaces and consolidate similar ones while independently modeling normal features within each space. Experiments on Anomaly-ShapeNet and Real3D-AD with different settings demonstrate that progressively finer-grained semantic alignment in preprocessing directly enhances point-level AD accuracy, providing inverse validation of the theorem's causal logic.
♻ ☆ AIR-HLoc: Adaptive Retrieved Images Selection for Efficient Visual Localisation ICRA
State-of-the-art hierarchical localisation pipelines (HLoc) employ image retrieval (IR) to establish 2D-3D correspondences by selecting the top-$k$ most similar images from a reference database. While increasing $k$ improves localisation robustness, it also linearly increases computational cost and runtime, creating a significant bottleneck. This paper investigates the relationship between global and local descriptors, showing that greater similarity between the global descriptors of query and database images increases the proportion of feature matches. Low similarity queries significantly benefit from increasing $k$, while high similarity queries rapidly experience diminishing returns. Building on these observations, we propose an adaptive strategy that adjusts $k$ based on the similarity between the query's global descriptor and those in the database, effectively mitigating the feature-matching bottleneck. Our approach optimizes processing time without sacrificing accuracy. Experiments on three indoor and outdoor datasets show that AIR-HLoc reduces feature matching time by up to 30\%, while preserving state-of-the-art accuracy. The results demonstrate that AIR-HLoc facilitates a latency-sensitive localisation system.
comment: Accepted to the 2025 IEEE International Conference on Robotics and Automation (ICRA)
♻ ☆ TweedieMix: Improving Multi-Concept Fusion for Diffusion-based Image/Video Generation
Despite significant advancements in customizing text-to-image and video generation models, generating images and videos that effectively integrate multiple personalized concepts remains a challenging task. To address this, we present TweedieMix, a novel method for composing customized diffusion models during the inference phase. By analyzing the properties of reverse diffusion sampling, our approach divides the sampling process into two stages. During the initial steps, we apply a multiple object-aware sampling technique to ensure the inclusion of the desired target objects. In the later steps, we blend the appearances of the custom concepts in the de-noised image space using Tweedie's formula. Our results demonstrate that TweedieMix can generate multiple personalized concepts with higher fidelity than existing methods. Moreover, our framework can be effortlessly extended to image-to-video diffusion models, enabling the generation of videos that feature multiple personalized concepts. Results and source code are in our anonymous project page.
comment: Github Page: https://github.com/KwonGihyun/TweedieMix
♻ ☆ LocoVR: Multiuser Indoor Locomotion Dataset in Virtual Reality ICLR2025
Understanding human locomotion is crucial for AI agents such as robots, particularly in complex indoor home environments. Modeling human trajectories in these spaces requires insight into how individuals maneuver around physical obstacles and manage social navigation dynamics. These dynamics include subtle behaviors influenced by proxemics - the social use of space, such as stepping aside to allow others to pass or choosing longer routes to avoid collisions. Previous research has developed datasets of human motion in indoor scenes, but these are often limited in scale and lack the nuanced social navigation dynamics common in home environments. To address this, we present LocoVR, a dataset of 7000+ two-person trajectories captured in virtual reality from over 130 different indoor home environments. LocoVR provides accurate trajectory data and precise spatial information, along with rich examples of socially-motivated movement behaviors. For example, the dataset captures instances of individuals navigating around each other in narrow spaces, adjusting paths to respect personal boundaries in living areas, and coordinating movements in high-traffic zones like entryways and kitchens. Our evaluation shows that LocoVR significantly enhances model performance in three practical indoor tasks utilizing human trajectories, and demonstrates predicting socially-aware navigation patterns in home environments.
comment: This paper has been accepted to ICLR2025
Machine Learning 150
☆ Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulation
Humans can accomplish complex contact-rich tasks using vision and touch, with highly reactive capabilities such as quick adjustments to environmental changes and adaptive control of contact forces; however, this remains challenging for robots. Existing visual imitation learning (IL) approaches rely on action chunking to model complex behaviors, which lacks the ability to respond instantly to real-time tactile feedback during the chunk execution. Furthermore, most teleoperation systems struggle to provide fine-grained tactile / force feedback, which limits the range of tasks that can be performed. To address these challenges, we introduce TactAR, a low-cost teleoperation system that provides real-time tactile feedback through Augmented Reality (AR), along with Reactive Diffusion Policy (RDP), a novel slow-fast visual-tactile imitation learning algorithm for learning contact-rich manipulation skills. RDP employs a two-level hierarchy: (1) a slow latent diffusion policy for predicting high-level action chunks in latent space at low frequency, (2) a fast asymmetric tokenizer for closed-loop tactile feedback control at high frequency. This design enables both complex trajectory modeling and quick reactive behavior within a unified framework. Through extensive evaluation across three challenging contact-rich tasks, RDP significantly improves performance compared to state-of-the-art visual IL baselines through rapid response to tactile / force feedback. Furthermore, experiments show that RDP is applicable across different tactile / force sensors. Code and videos are available on https://reactive-diffusion-policy.github.io/.
☆ A New $\sim 5σ$ Tension at Characteristic Redshift from DESI DR1 and DES-SN5YR observations
We perform a model-independent reconstruction of the angular diameter distance ($D_{A}$) using the Multi-Task Gaussian Process (MTGP) framework with DESI-DR1 BAO and DES-SN5YR datasets. We calibrate the comoving sound horizon at the baryon drag epoch $r_d$ to the Planck best-fit value, ensuring consistency with early-universe physics. With the reconstructed $D_A$ at two key redshifts, $z\sim 1.63$ (where $D_{A}^{\prime} =0$) and at $z\sim 0.512$ (where $D_{A}^{\prime} = D_{A}$), we derive the expansion rate of the Universe $H(z)$ at these redshifts. Our findings reveal that at $z\sim 1.63$, the $H(z)$ is fully consistent with the Planck-2018 $\Lambda$CDM prediction, confirming no new physics at that redshift. However, at $z \sim 0.512$, the derived $H(z)$ shows a more than $5\sigma$ discrepancy with the Planck-2018 $\Lambda$CDM prediction, suggesting a possible breakdown of the $\Lambda$CDM model as constrained by Planck-2018 at this lower redshift. This emerging $\sim 5\sigma$ tension at $z\sim 0.512$, distinct from the existing ``Hubble Tension'', may signal the first strong evidence for new physics at low redshifts.
comment: 4 pages, 1 table, 1 figure. Comments are welcome
☆ Wikipedia in the Era of LLMs: Evolution and Risks
In this paper, we present a thorough analysis of the impact of Large Language Models (LLMs) on Wikipedia, examining the evolution of Wikipedia through existing data and using simulations to explore potential risks. We begin by analyzing page views and article content to study Wikipedia's recent changes and assess the impact of LLMs. Subsequently, we evaluate how LLMs affect various Natural Language Processing (NLP) tasks related to Wikipedia, including machine translation and retrieval-augmented generation (RAG). Our findings and simulation results reveal that Wikipedia articles have been influenced by LLMs, with an impact of approximately 1%-2% in certain categories. If the machine translation benchmark based on Wikipedia is influenced by LLMs, the scores of the models may become inflated, and the comparative results among models might shift as well. Moreover, the effectiveness of RAG might decrease if the knowledge base becomes polluted by LLM-generated content. While LLMs have not yet fully changed Wikipedia's language and knowledge structures, we believe that our empirical findings signal the need for careful consideration of potential future risks.
comment: We release all the experimental dataset and source code at: https://github.com/HSM316/LLM_Wikipedia
☆ Language Models can Self-Improve at State-Value Estimation for Better Search
Collecting ground truth task completion rewards or human demonstrations for multi-step reasoning tasks is often cost-prohibitive and time-consuming, especially in interactive domains like web tasks. To address this bottleneck, we present self-taught lookahead, a self-supervised method that leverages state-transition dynamics to train a value model capable of effectively guiding language model-controlled search. We find that moderately sized (8 billion parameters) open-weight value models improved with self-taught lookahead can match the performance of using a frontier LLM such as gpt-4o as the value model. Furthermore, we find that self-taught lookahead improves performance by 20% while reducing costs 37x compared to previous LLM-based tree search, without relying on ground truth rewards.
☆ Weak-to-Strong Generalization Even in Random Feature Networks, Provably
Weak-to-Strong Generalization (Burns et al., 2024) is the phenomenon whereby a strong student, say GPT-4, learns a task from a weak teacher, say GPT-2, and ends up significantly outperforming the teacher. We show that this phenomenon does not require a strong learner like GPT-4. We consider student and teacher that are random feature models, described by two-layer networks with a random and fixed bottom layer and a trained top layer. A "weak" teacher, with a small number of units (i.e. random features), is trained on the population, and a "strong" student, with a much larger number of units (i.e. random features), is trained only on labels generated by the weak teacher. We demonstrate, prove, and understand how the student can outperform the teacher, even though trained only on data labeled by the teacher. We also explain how such weak-to-strong generalization is enabled by early stopping. Importantly, we also show the quantitative limits of weak-to-strong generalization in this model.
☆ Multiaccuracy and Multicalibration via Proxy Groups
As the use of predictive machine learning algorithms increases in high-stakes decision-making, it is imperative that these algorithms are fair across sensitive groups. Unfortunately, measuring and enforcing fairness in real-world applications can be challenging due to missing or incomplete sensitive group data. Proxy-sensitive attributes have been proposed as a practical and effective solution in these settings, but only for parity-based fairness notions. Knowing how to evaluate and control for fairness with missing sensitive group data for newer and more flexible frameworks, such as multiaccuracy and multicalibration, remains unexplored. In this work, we address this gap by demonstrating that in the absence of sensitive group data, proxy-sensitive attributes can provably be used to derive actionable upper bounds on the true multiaccuracy and multicalibration, providing insights into a model's potential worst-case fairness violations. Additionally, we show that adjusting models to satisfy multiaccuracy and multicalibration across proxy-sensitive attributes can significantly mitigate these violations for the true, but unknown, sensitive groups. Through several experiments on real-world datasets, we illustrate that approximate multiaccuracy and multicalibration can be achieved even when sensitive group information is incomplete or unavailable.
☆ Calibrating LLM Confidence with Semantic Steering: A Multi-Prompt Aggregation Framework
Large Language Models (LLMs) often exhibit misaligned confidence scores, usually overestimating the reliability of their predictions. While verbalized confidence in Large Language Models (LLMs) has gained attention, prior work remains divided on whether confidence scores can be systematically steered through prompting. Recent studies even argue that such prompt-induced confidence shifts are negligible, suggesting LLMs' confidence calibration is rigid to linguistic interventions. Contrary to these claims, we first rigorously confirm the existence of directional confidence shifts by probing three models (including GPT3.5, LLAMA3-70b, GPT4) across 7 benchmarks, demonstrating that explicit instructions can inflate or deflate confidence scores in a regulated manner. Based on this observation, we propose a novel framework containing three components: confidence steering, steered confidence aggregation and steered answers selection, named SteeringConf. Our method, SteeringConf, leverages a confidence manipulation mechanism to steer the confidence scores of LLMs in several desired directions, followed by a summarization module that aggregates the steered confidence scores to produce a final prediction. We evaluate our method on 7 benchmarks and it consistently outperforms the baselines in terms of calibration metrics in task of confidence calibration and failure detection.
☆ Unsupervised Attributed Dynamic Network Embedding with Stability Guarantees
Stability for dynamic network embeddings ensures that nodes behaving the same at different times receive the same embedding, allowing comparison of nodes in the network across time. We present attributed unfolded adjacency spectral embedding (AUASE), a stable unsupervised representation learning framework for dynamic networks in which nodes are attributed with time-varying covariate information. To establish stability, we prove uniform convergence to an associated latent position model. We quantify the benefits of our dynamic embedding by comparing with state-of-the-art network representation learning methods on three real attributed networks. To the best of our knowledge, AUASE is the only attributed dynamic embedding that satisfies stability guarantees without the need for ground truth labels, which we demonstrate provides significant improvements for link prediction and node classification.
comment: 27 pages, 5 figures
☆ (How) Do Language Models Track State?
Transformer language models (LMs) exhibit behaviors -- from storytelling to code generation -- that appear to require tracking the unobserved state of an evolving world. How do they do so? We study state tracking in LMs trained or fine-tuned to compose permutations (i.e., to compute the order of a set of objects after a sequence of swaps). Despite the simple algebraic structure of this problem, many other tasks (e.g., simulation of finite automata and evaluation of boolean expressions) can be reduced to permutation composition, making it a natural model for state tracking in general. We show that LMs consistently learn one of two state tracking mechanisms for this task. The first closely resembles the "associative scan" construction used in recent theoretical work by Liu et al. (2023) and Merrill et al. (2024). The second uses an easy-to-compute feature (permutation parity) to partially prune the space of outputs, then refines this with an associative scan. The two mechanisms exhibit markedly different robustness properties, and we show how to steer LMs toward one or the other with intermediate training tasks that encourage or suppress the heuristics. Our results demonstrate that transformer LMs, whether pretrained or fine-tuned, can learn to implement efficient and interpretable state tracking mechanisms, and the emergence of these mechanisms can be predicted and controlled.
comment: 21 pages, 17 figures, 1 table
☆ Beyond Cosine Decay: On the effectiveness of Infinite Learning Rate Schedule for Continual Pre-training
The ever-growing availability of unlabeled data presents both opportunities and challenges for training artificial intelligence systems. While self-supervised learning (SSL) has emerged as a powerful paradigm for extracting meaningful representations from vast amounts of unlabeled data, existing methods still struggle to adapt to the non-stationary, non-IID nature of real-world data streams without forgetting previously learned knowledge. Recent works have adopted a repeated cosine annealing schedule for large-scale continual pre-training; however, these schedules (1) inherently cause forgetting during the re-warming phase and (2) have not been systematically compared to existing continual SSL methods. In this work, we systematically compare the widely used cosine schedule with the recently proposed infinite learning rate schedule and empirically find the latter to be a more effective alternative. Our extensive empirical evaluation across diverse image and language datasets demonstrates that the infinite learning rate schedule consistently enhances continual pre-training performance compared to a repeated cosine decay without being restricted to a fixed iteration budget. For instance, in a small-scale MAE pre-training setup, it outperforms several strong baselines from the literature. We then scale up our experiments to larger MAE pre-training and autoregressive language model pre-training. Our results show that the infinite learning rate schedule remains effective at scale, surpassing repeated cosine decay for both MAE pre-training and zero-shot LM benchmarks.
☆ SeqFusion: Sequential Fusion of Pre-Trained Models for Zero-Shot Time-Series Forecasting
Unlike traditional time-series forecasting methods that require extensive in-task data for training, zero-shot forecasting can directly predict future values given a target time series without additional training data. Current zero-shot approaches primarily rely on pre-trained generalized models, with their performance often depending on the variety and relevance of the pre-training data, which can raise privacy concerns. Instead of collecting diverse pre-training data, we introduce SeqFusion in this work, a novel framework that collects and fuses diverse pre-trained models (PTMs) sequentially for zero-shot forecasting. Based on the specific temporal characteristics of the target time series, SeqFusion selects the most suitable PTMs from a batch of pre-collected PTMs, performs sequential predictions, and fuses all the predictions while using minimal data to protect privacy. Each of these PTMs specializes in different temporal patterns and forecasting tasks, allowing SeqFusion to select by measuring distances in a shared representation space of the target time series with each PTM. Experiments demonstrate that SeqFusion achieves competitive accuracy in zero-shot forecasting compared to state-of-the-art methods.
☆ AlignDistil: Token-Level Language Model Alignment as Adaptive Policy Distillation
In modern large language models (LLMs), LLM alignment is of crucial importance and is typically achieved through methods such as reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO). However, in most existing methods for LLM alignment, all tokens in the response are optimized using a sparse, response-level reward or preference annotation. The ignorance of token-level rewards may erroneously punish high-quality tokens or encourage low-quality tokens, resulting in suboptimal performance and slow convergence speed. To address this issue, we propose AlignDistil, an RLHF-equivalent distillation method for token-level reward optimization. Specifically, we introduce the reward learned by DPO into the RLHF objective and theoretically prove the equivalence between this objective and a token-level distillation process, where the teacher distribution linearly combines the logits from the DPO model and a reference model. On this basis, we further bridge the accuracy gap between the reward from the DPO model and the pure reward model, by building a contrastive DPO reward with a normal and a reverse DPO model. Moreover, to avoid under- and over-optimization on different tokens, we design a token adaptive logit extrapolation mechanism to construct an appropriate teacher distribution for each token. Experimental results demonstrate the superiority of our AlignDistil over existing methods and showcase fast convergence due to its token-level distributional reward optimization.
comment: 15 pages, 2 figures
☆ Meta-Learning to Explore via Memory Density Feedback
Exploration algorithms for reinforcement learning typically replace or augment the reward function with an additional ``intrinsic'' reward that trains the agent to seek previously unseen states of the environment. Here, we consider an exploration algorithm that exploits meta-learning, or learning to learn, such that the agent learns to maximize its exploration progress within a single episode, even between epochs of training. The agent learns a policy that aims to minimize the probability density of new observations with respect to all of its memories. In addition, it receives as feedback evaluations of the current observation density and retains that feedback in a recurrent network. By remembering trajectories of density, the agent learns to navigate a complex and growing landscape of familiarity in real-time, allowing it to maximize its exploration progress even in completely novel states of the environment for which its policy has not been trained.
comment: 15 pages, 6 figures
☆ On Separation Between Best-Iterate, Random-Iterate, and Last-Iterate Convergence of Learning in Games
Non-ergodic convergence of learning dynamics in games is widely studied recently because of its importance in both theory and practice. Recent work (Cai et al., 2024) showed that a broad class of learning dynamics, including Optimistic Multiplicative Weights Update (OMWU), can exhibit arbitrarily slow last-iterate convergence even in simple $2 \times 2$ matrix games, despite many of these dynamics being known to converge asymptotically in the last iterate. It remains unclear, however, whether these algorithms achieve fast non-ergodic convergence under weaker criteria, such as best-iterate convergence. We show that for $2\times 2$ matrix games, OMWU achieves an $O(T^{-1/6})$ best-iterate convergence rate, in stark contrast to its slow last-iterate convergence in the same class of games. Furthermore, we establish a lower bound showing that OMWU does not achieve any polynomial random-iterate convergence rate, measured by the expected duality gaps across all iterates. This result challenges the conventional wisdom that random-iterate convergence is essentially equivalent to best-iterate convergence, with the former often used as a proxy for establishing the latter. Our analysis uncovers a new connection to dynamic regret and presents a novel two-phase approach to best-iterate convergence, which could be of independent interest.
comment: 33 pages
☆ Feynman-Kac Correctors in Diffusion: Annealing, Guidance, and Product of Experts
While score-based generative models are the model of choice across diverse domains, there are limited tools available for controlling inference-time behavior in a principled manner, e.g. for composing multiple pretrained models. Existing classifier-free guidance methods use a simple heuristic to mix conditional and unconditional scores to approximately sample from conditional distributions. However, such methods do not approximate the intermediate distributions, necessitating additional 'corrector' steps. In this work, we provide an efficient and principled method for sampling from a sequence of annealed, geometric-averaged, or product distributions derived from pretrained score-based models. We derive a weighted simulation scheme which we call Feynman-Kac Correctors (FKCs) based on the celebrated Feynman-Kac formula by carefully accounting for terms in the appropriate partial differential equations (PDEs). To simulate these PDEs, we propose Sequential Monte Carlo (SMC) resampling algorithms that leverage inference-time scaling to improve sampling quality. We empirically demonstrate the utility of our methods by proposing amortized sampling via inference-time temperature annealing, improving multi-objective molecule generation using pretrained models, and improving classifier-free guidance for text-to-image generation. Our code is available at https://github.com/martaskrt/fkc-diffusion.
☆ A Minimalist Example of Edge-of-Stability and Progressive Sharpening
Recent advances in deep learning optimization have unveiled two intriguing phenomena under large learning rates: Edge of Stability (EoS) and Progressive Sharpening (PS), challenging classical Gradient Descent (GD) analyses. Current research approaches, using either generalist frameworks or minimalist examples, face significant limitations in explaining these phenomena. This paper advances the minimalist approach by introducing a two-layer network with a two-dimensional input, where one dimension is relevant to the response and the other is irrelevant. Through this model, we rigorously prove the existence of progressive sharpening and self-stabilization under large learning rates, and establish non-asymptotic analysis of the training dynamics and sharpness along the entire GD trajectory. Besides, we connect our minimalist example to existing works by reconciling the existence of a well-behaved ``stable set" between minimalist and generalist analyses, and extending the analysis of Gradient Flow Solution sharpness to our two-dimensional input scenario. These findings provide new insights into the EoS phenomenon from both parameter and input data distribution perspectives, potentially informing more effective optimization strategies in deep learning practice.
comment: 39 pages, 15 figures
☆ Inductive randomness predictors
This paper introduces inductive randomness predictors, which form a superset of inductive conformal predictors. Its focus is on a very simple special case, binary inductive randomness predictors. It is interesting that binary inductive randomness predictors have an advantage over inductive conformal predictors, although they also have a serious disadvantage. This advantage will allow us to reach the surprising conclusion that non-trivial inductive conformal predictors are inadmissible in the sense of statistical decision theory.
comment: 10 pages, 1 table
☆ RAAD-LLM: Adaptive Anomaly Detection Using LLMs and RAG Integration
Anomaly detection in complex industrial environments poses unique challenges, particularly in contexts characterized by data sparsity and evolving operational conditions. Predictive maintenance (PdM) in such settings demands methodologies that are adaptive, transferable, and capable of integrating domain-specific knowledge. In this paper, we present RAAD-LLM, a novel framework for adaptive anomaly detection, leveraging large language models (LLMs) integrated with Retrieval-Augmented Generation (RAG). This approach addresses the aforementioned PdM challenges. By effectively utilizing domain-specific knowledge, RAAD-LLM enhances the detection of anomalies in time series data without requiring fine-tuning on specific datasets. The framework's adaptability mechanism enables it to adjust its understanding of normal operating conditions dynamically, thus increasing detection accuracy. We validate this methodology through a real-world application for a plastics manufacturing plant and the Skoltech Anomaly Benchmark (SKAB). Results show significant improvements over our previous model with an accuracy increase from 70.7 to 89.1 on the real-world dataset. By allowing for the enriching of input series data with semantics, RAAD-LLM incorporates multimodal capabilities that facilitate more collaborative decision-making between the model and plant operators. Overall, our findings support RAAD-LLM's ability to revolutionize anomaly detection methodologies in PdM, potentially leading to a paradigm shift in how anomaly detection is implemented across various industries.
comment: arXiv admin note: substantial text overlap with arXiv:2411.00914
☆ Spike-and-Slab Posterior Sampling in High Dimensions
Posterior sampling with the spike-and-slab prior [MB88], a popular multimodal distribution used to model uncertainty in variable selection, is considered the theoretical gold standard method for Bayesian sparse linear regression [CPS09, Roc18]. However, designing provable algorithms for performing this sampling task is notoriously challenging. Existing posterior samplers for Bayesian sparse variable selection tasks either require strong assumptions about the signal-to-noise ratio (SNR) [YWJ16], only work when the measurement count grows at least linearly in the dimension [MW24], or rely on heuristic approximations to the posterior. We give the first provable algorithms for spike-and-slab posterior sampling that apply for any SNR, and use a measurement count sublinear in the problem dimension. Concretely, assume we are given a measurement matrix $\mathbf{X} \in \mathbb{R}^{n\times d}$ and noisy observations $\mathbf{y} = \mathbf{X}\mathbf{\theta}^\star + \mathbf{\xi}$ of a signal $\mathbf{\theta}^\star$ drawn from a spike-and-slab prior $\pi$ with a Gaussian diffuse density and expected sparsity k, where $\mathbf{\xi} \sim \mathcal{N}(\mathbb{0}_n, \sigma^2\mathbf{I}_n)$. We give a polynomial-time high-accuracy sampler for the posterior $\pi(\cdot \mid \mathbf{X}, \mathbf{y})$, for any SNR $\sigma^{-1}$ > 0, as long as $n \geq k^3 \cdot \text{polylog}(d)$ and $X$ is drawn from a matrix ensemble satisfying the restricted isometry property. We further give a sampler that runs in near-linear time $\approx nd$ in the same setting, as long as $n \geq k^5 \cdot \text{polylog}(d)$. To demonstrate the flexibility of our framework, we extend our result to spike-and-slab posterior sampling with Laplace diffuse densities, achieving similar guarantees when $\sigma = O(\frac{1}{k})$ is bounded.
comment: 53 pages
☆ IterPref: Focal Preference Learning for Code Generation via Iterative Debugging
Preference learning enhances Code LLMs beyond supervised fine-tuning by leveraging relative quality comparisons. Existing methods construct preference pairs from candidates based on test case success, treating the higher pass rate sample as positive and the lower as negative. However, this approach does not pinpoint specific errors in the code, which prevents the model from learning more informative error correction patterns, as aligning failing code as a whole lacks the granularity needed to capture meaningful error-resolution relationships. To address these issues, we propose IterPref, a new preference alignment framework that mimics human iterative debugging to refine Code LLMs. IterPref explicitly locates error regions and aligns the corresponding tokens via a tailored DPO algorithm. To generate informative pairs, we introduce the CodeFlow dataset, where samples are iteratively refined until passing tests, with modifications capturing error corrections. Extensive experiments show that a diverse suite of Code LLMs equipped with IterPref achieves significant performance gains in code generation and improves on challenging tasks like BigCodeBench. In-depth analysis reveals that IterPref yields fewer errors. Our code and data will be made publicaly available.
comment: The code and data will be released soon
☆ Quantitative Resilience Modeling for Autonomous Cyber Defense
Cyber resilience is the ability of a system to recover from an attack with minimal impact on system operations. However, characterizing a network's resilience under a cyber attack is challenging, as there are no formal definitions of resilience applicable to diverse network topologies and attack patterns. In this work, we propose a quantifiable formulation of resilience that considers multiple defender operational goals, the criticality of various network resources for daily operations, and provides interpretability to security operators about their system's resilience under attack. We evaluate our approach within the CybORG environment, a reinforcement learning (RL) framework for autonomous cyber defense, analyzing trade-offs between resilience, costs, and prioritization of operational goals. Furthermore, we introduce methods to aggregate resilience metrics across time-variable attack patterns and multiple network topologies, comprehensively characterizing system resilience. Using insights gained from our resilience metrics, we design RL autonomous defensive agents and compare them against several heuristic baselines, showing that proactive network hardening techniques and prompt recovery of compromised machines are critical for effective cyber defenses.
☆ Prime Convolutional Model: Breaking the Ground for Theoretical Explainability
In this paper, we propose a new theoretical approach to Explainable AI. Following the Scientific Method, this approach consists in formulating on the basis of empirical evidence, a mathematical model to explain and predict the behaviors of Neural Networks. We apply the method to a case study created in a controlled environment, which we call Prime Convolutional Model (p-Conv for short). p-Conv operates on a dataset consisting of the first one million natural numbers and is trained to identify the congruence classes modulo a given integer $m$. Its architecture uses a convolutional-type neural network that contextually processes a sequence of $B$ consecutive numbers to each input. We take an empirical approach and exploit p-Conv to identify the congruence classes of numbers in a validation set using different values for $m$ and $B$. The results show that the different behaviors of p-Conv (i.e., whether it can perform the task or not) can be modeled mathematically in terms of $m$ and $B$. The inferred mathematical model reveals interesting patterns able to explain when and why p-Conv succeeds in performing task and, if not, which error pattern it follows.
☆ Efficient and Optimal No-Regret Caching under Partial Observation
Online learning algorithms have been successfully used to design caching policies with sublinear regret in the total number of requests, with no statistical assumption about the request sequence. Most existing algorithms involve computationally expensive operations and require knowledge of all past requests. However, this may not be feasible in practical scenarios like caching at a cellular base station. Therefore, we study the caching problem in a more restrictive setting where only a fraction of past requests are observed, and we propose a randomized caching policy with sublinear regret based on the classic online learning algorithm Follow-the-Perturbed-Leader (FPL). Our caching policy is the first to attain the asymptotically optimal regret bound while ensuring asymptotically constant amortized time complexity in the partial observability setting of requests. The experimental evaluation compares the proposed solution against classic caching policies and validates the proposed approach under synthetic and real-world request traces.
☆ ArcPro: Architectural Programs for Structured 3D Abstraction of Sparse Points CVPR 2025
We introduce ArcPro, a novel learning framework built on architectural programs to recover structured 3D abstractions from highly sparse and low-quality point clouds. Specifically, we design a domain-specific language (DSL) to hierarchically represent building structures as a program, which can be efficiently converted into a mesh. We bridge feedforward and inverse procedural modeling by using a feedforward process for training data synthesis, allowing the network to make reverse predictions. We train an encoder-decoder on the points-program pairs to establish a mapping from unstructured point clouds to architectural programs, where a 3D convolutional encoder extracts point cloud features and a transformer decoder autoregressively predicts the programs in a tokenized form. Inference by our method is highly efficient and produces plausible and faithful 3D abstractions. Comprehensive experiments demonstrate that ArcPro outperforms both traditional architectural proxy reconstruction and learning-based abstraction methods. We further explore its potential to work with multi-view image and natural language inputs.
comment: CVPR 2025 (Patent Protected); Project page: https://vcc.tech/research/2025/ArcPro
☆ Seeded Poisson Factorization: Leveraging domain knowledge to fit topic models
Topic models are widely used for discovering latent thematic structures in large text corpora, yet traditional unsupervised methods often struggle to align with predefined conceptual domains. This paper introduces Seeded Poisson Factorization (SPF), a novel approach that extends the Poisson Factorization framework by incorporating domain knowledge through seed words. SPF enables a more interpretable and structured topic discovery by modifying the prior distribution of topic-specific term intensities, assigning higher initial rates to predefined seed words. The model is estimated using variational inference with stochastic gradient optimization, ensuring scalability to large datasets. We apply SPF to an Amazon customer feedback dataset, leveraging predefined product categories as guiding structures. Our evaluation demonstrates that SPF achieves superior classification performance compared to alternative guided topic models, particularly in terms of computational efficiency and predictive performance. Furthermore, robustness checks highlight SPF's ability to adaptively balance domain knowledge and data-driven topic discovery, even in cases of imperfect seed word selection. These results establish SPF as a powerful and scalable alternative for integrating expert knowledge into topic modeling, enhancing both interpretability and efficiency in real-world applications.
☆ Clustered KL-barycenter design for policy evaluation
In the context of stochastic bandit models, this article examines how to design sample-efficient behavior policies for the importance sampling evaluation of multiple target policies. From importance sampling theory, it is well established that sample efficiency is highly sensitive to the KL divergence between the target and importance sampling distributions. We first analyze a single behavior policy defined as the KL-barycenter of the target policies. Then, we refine this approach by clustering the target policies into groups with small KL divergences and assigning each cluster its own KL-barycenter as a behavior policy. This clustered KL-based policy evaluation (CKL-PE) algorithm provides a novel perspective on optimal policy selection. We prove upper bounds on the sample complexity of our method and demonstrate its effectiveness with numerical validation.
☆ S4D-Bio Audio Monitoring of Bone Cement Disintegration in Pulsating Fluid Jet Surgery under Laboratory Conditions
This study investigates a pulsating fluid jet as a novel precise, minimally invasive and cold technique for bone cement removal. We utilize the pulsating fluid jet device to remove bone cement from samples designed to mimic clinical conditions. The effectiveness of long nozzles was tested to enable minimally invasive procedures. Audio signal monitoring, complemented by the State Space Model (SSM) S4D-Bio, was employed to optimize the fluid jet parameters dynamically, addressing challenges like visibility obstruction from splashing. Within our experiments, we generate a comprehensive dataset correlating various process parameters and their equivalent audio signals to material erosion. The use of SSMs yields precise control over the predictive erosion process, achieving 98.93 \% accuracy. The study demonstrates on the one hand, that the pulsating fluid jet device, coupled with advanced audio monitoring techniques, is a highly effective tool for precise bone cement removal. On the other hand, this study presents the first application of SSMs in biomedical surgery technology, marking a significant advancement in the application. This research significantly advances biomedical engineering by integrating machine learning combined with pulsating fluid jet as surgical technology, offering a novel, minimally invasive, cold and adaptive approach for bone cement removal in orthopedic applications.
comment: submitted to Computers in Biology and Medicine Journal
☆ RedChronos: A Large Language Model-Based Log Analysis System for Insider Threat Detection in Enterprises
Internal threat detection aims to address security threats within organizations or enterprises by identifying potential or already occurring malicious threats within vast amounts of logs. Although organizations or enterprises have dedicated personnel responsible for reviewing these logs, it is impossible to manually examine all logs entirely. In response to the vast number of logs, we propose a system called RedChronos, which is a Large Language Model-Based Log Analysis System. This system incorporates innovative improvements over previous research by employing Query-Aware Weighted Voting and a Semantic Expansion-based Genetic Algorithm with LLM-driven Mutations. On the public datasets CERT 4.2 and 5.2, RedChronos outperforms or matches existing approaches in terms of accuracy, precision, and detection rate. Moreover, RedChronos reduces the need for manual intervention in security log reviews by 90\% in the Xiaohongshu SOC. Therefore, our RedChronos system demonstrates exceptional performance in handling Internal Threat Detection (IDT) tasks, providing innovative solutions for these challenges. We believe that future research can continue to enhance the system's performance in IDT tasks while also reducing the response time to internal risk events.
☆ Federated Learning for Privacy-Preserving Feedforward Control in Multi-Agent Systems IJCNN 2025
Feedforward control (FF) is often combined with feedback control (FB) in many control systems, improving tracking performance, efficiency, and stability. However, designing effective data-driven FF controllers in multi-agent systems requires significant data collection, including transferring private or proprietary data, which raises privacy concerns and incurs high communication costs. Therefore, we propose a novel approach integrating Federated Learning (FL) into FF control to address these challenges. This approach enables privacy-preserving, communication-efficient, and decentralized continuous improvement of FF controllers across multiple agents without sharing personal or proprietary data. By leveraging FL, each agent learns a local, neural FF controller using its data and contributes only model updates to a global aggregation process, ensuring data privacy and scalability. We demonstrate the effectiveness of our method in an autonomous driving use case. Therein, vehicles equipped with a trajectory-tracking feedback controller are enhanced by FL-based neural FF control. Simulations highlight significant improvements in tracking performance compared to pure FB control, analogous to model-based FF control. We achieve comparable tracking performance without exchanging private vehicle-specific data compared to a centralized neural FF control. Our results underscore the potential of FL-based neural FF control to enable privacy-preserving learning in multi-agent control systems, paving the way for scalable and efficient autonomous systems applications.
comment: Submitted to IJCNN 2025
☆ Generative Modeling of Microweather Wind Velocities for Urban Air Mobility
Motivated by the pursuit of safe, reliable, and weather-tolerant urban air mobility (UAM) solutions, this work proposes a generative modeling approach for characterizing microweather wind velocities. Microweather, or the weather conditions in highly localized areas, is particularly complex in urban environments owing to the chaotic and turbulent nature of wind flows. Furthermore, traditional means of assessing local wind fields are not generally viable solutions for UAM applications: 1) field measurements that would rely on permanent wind profiling systems in operational air space are not practical, 2) physics-based models that simulate fluid dynamics at a sufficiently high resolution are not computationally tractable, and 3) data-driven modeling approaches that are largely deterministic ignore the inherent variability in turbulent flows that dictates UAM reliability. Thus, advancements in predictive capabilities are needed to help mitigate the unique operational safety risks that microweather winds pose for smaller, lighter weight UAM aircraft. This work aims to model microweather wind velocities in a manner that is computationally-efficient, captures random variability, and would only require a temporary, rather than permanent, field measurement campaign. Inspired by recent breakthroughs in conditional generative AI such as text-to-image generation, the proposed approach learns a probabilistic macro-to-microweather mapping between regional weather forecasts and measured local wind velocities using generative modeling (denoising diffusion probabilistic models, flow matching, and Gaussian mixture models). A simple proof of concept was implemented using a dataset comprised of local (micro) measurements from a Sonic Detection and Ranging (SoDAR) wind profiler along with (macro) forecast data from a nearby weather station over the same time period.
comment: 17 pages, 13 figures, published in 2025 IEEE Aerospace Conference proceedings
☆ Class-Aware PillarMix: Can Mixed Sample Data Augmentation Enhance 3D Object Detection with Radar Point Clouds? IROS 2025
Due to the significant effort required for data collection and annotation in 3D perception tasks, mixed sample data augmentation (MSDA) has been widely studied to generate diverse training samples by mixing existing data. Recently, many MSDA techniques have been developed for point clouds, but they mainly target LiDAR data, leaving their application to radar point clouds largely unexplored. In this paper, we examine the feasibility of applying existing MSDA methods to radar point clouds and identify several challenges in adapting these techniques. These obstacles stem from the radar's irregular angular distribution, deviations from a single-sensor polar layout in multi-radar setups, and point sparsity. To address these issues, we propose Class-Aware PillarMix (CAPMix), a novel MSDA approach that applies MixUp at the pillar level in 3D point clouds, guided by class labels. Unlike methods that rely a single mix ratio to the entire sample, CAPMix assigns an independent ratio to each pillar, boosting sample diversity. To account for the density of different classes, we use class-specific distributions: for dense objects (e.g., large vehicles), we skew ratios to favor points from another sample, while for sparse objects (e.g., pedestrians), we sample more points from the original. This class-aware mixing retains critical details and enriches each sample with new information, ultimately generating more diverse training data. Experimental results demonstrate that our method not only significantly boosts performance but also outperforms existing MSDA approaches across two datasets (Bosch Street and K-Radar). We believe that this straightforward yet effective approach will spark further investigation into MSDA techniques for radar data.
comment: 8 pages, 6 figures, 4 tables, submitted to 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025)
☆ MPO: Boosting LLM Agents with Meta Plan Optimization
Recent advancements in large language models (LLMs) have enabled LLM-based agents to successfully tackle interactive planning tasks. However, despite their successes, existing approaches often suffer from planning hallucinations and require retraining for each new agent. To address these challenges, we propose the Meta Plan Optimization (MPO) framework, which enhances agent planning capabilities by directly incorporating explicit guidance. Unlike previous methods that rely on complex knowledge, which either require significant human effort or lack quality assurance, MPO leverages high-level general guidance through meta plans to assist agent planning and enables continuous optimization of the meta plans based on feedback from the agent's task execution. Our experiments conducted on two representative tasks demonstrate that MPO significantly outperforms existing baselines. Moreover, our analysis indicates that MPO provides a plug-and-play solution that enhances both task completion efficiency and generalization capabilities in previous unseen scenarios.
☆ VWAP Execution with Signature-Enhanced Transformers: A Multi-Asset Learning Approach
In this paper I propose a novel approach to Volume Weighted Average Price (VWAP) execution that addresses two key practical challenges: the need for asset-specific model training and the capture of complex temporal dependencies. Building upon my recent work in dynamic VWAP execution arXiv:2502.18177, I demonstrate that a single neural network trained across multiple assets can achieve performance comparable to or better than traditional asset-specific models. The proposed architecture combines a transformer-based design inspired by arXiv:2406.02486 with path signatures for capturing geometric features of price-volume trajectories, as in arXiv:2406.17890. The empirical analysis, conducted on hourly cryptocurrency trading data from 80 trading pairs, shows that the globally-fitted model with signature features (GFT-Sig) achieves superior performance in both absolute and quadratic VWAP loss metrics compared to asset-specific approaches. Notably, these improvements persist for out-of-sample assets, demonstrating the model's ability to generalize across different market conditions. The results suggest that combining global parameter sharing with signature-based feature extraction provides a scalable and robust approach to VWAP execution, offering significant practical advantages over traditional asset-specific implementations.
☆ Weakly-Constrained 4D Var for Downscaling with Uncertainty using Data-Driven Surrogate Models
Dynamic downscaling typically involves using numerical weather prediction (NWP) solvers to refine coarse data to higher spatial resolutions. Data-driven models such as FourCastNet have emerged as a promising alternative to the traditional NWP models for forecasting. Once these models are trained, they are capable of delivering forecasts in a few seconds, thousands of times faster compared to classical NWP models. However, as the lead times, and, therefore, their forecast window, increase, these models show instability in that they tend to diverge from reality. In this paper, we propose to use data assimilation approaches to stabilize them when used for downscaling tasks. Data assimilation uses information from three different sources, namely an imperfect computational model based on partial differential equations (PDE), from noisy observations, and from an uncertainty-reflecting prior. In this work, when carrying out dynamic downscaling, we replace the computationally expensive PDE-based NWP models with FourCastNet in a ``weak-constrained 4DVar framework" that accounts for the implied model errors. We demonstrate the efficacy of this approach for a hurricane-tracking problem; moreover, the 4DVar framework naturally allows the expression and quantification of uncertainty. We demonstrate, using ERA5 data, that our approach performs better than the ensemble Kalman filter (EnKF) and the unstabilized FourCastNet model, both in terms of forecast accuracy and forecast uncertainty.
☆ Adapting Decoder-Based Language Models for Diverse Encoder Downstream Tasks
Decoder-based transformers, while revolutionizing language modeling and scaling to immense sizes, have not completely overtaken encoder-heavy architectures in natural language processing. Specifically, encoder-only models remain dominant in tasks like classification, regression, and ranking. This is primarily due to the inherent structure of decoder-based models, which limits their direct applicability to these tasks. In this paper, we introduce Gemma Encoder, adapting the powerful Gemma decoder model to an encoder architecture, thereby unlocking its potential for a wider range of non-generative applications. To optimize the adaptation from decoder to encoder, we systematically analyze various pooling strategies, attention mechanisms, and hyperparameters (e.g., dropout rate). Furthermore, we benchmark Gemma Encoder against established approaches on the GLUE benchmarks, and MS MARCO ranking benchmark, demonstrating its effectiveness and versatility.
☆ Cellular Automaton With CNN
Cellular automata (CA) models are widely used to simulate complex systems with emergent behaviors, but identifying hidden parameters that govern their dynamics remains a significant challenge. This study explores the use of Convolutional Neural Networks (CNN) to identify jump parameters in a two-dimensional CA model. We propose a custom CNN architecture trained on CA-generated data to classify jump parameters, which dictates the neighborhood size and movement rules of cells within the CA. Experiments were conducted across varying domain sizes (25 x 25 to 150 x 150) and CA iterations (0 to 50), demonstrating that the accuracy improves with larger domain sizes, as they provide more spatial information for parameter estimation. Interestingly, while initial CA iterations enhance the performance, increasing the number of iterations beyond a certain threshold does not significantly improve accuracy, suggesting that only specific temporal information is relevant for parameter identification. The proposed CNN achieves competitive accuracy (89.31) compared to established architectures like LeNet-5 and AlexNet, while offering significantly faster inference times, making it suitable for real-time applications. This study highlights the potential of CNNs as a powerful tool for fast and accurate parameter estimation in CA models, paving the way for their use in more complex systems and higher-dimensional domains. Future work will explore the identification of multiple hidden parameters and extend the approach to three-dimensional CA models.
☆ Weight transport through spike timing for robust local gradients
In both machine learning and in computational neuroscience, plasticity in functional neural networks is frequently expressed as gradient descent on a cost. Often, this imposes symmetry constraints that are difficult to reconcile with local computation, as is required for biological networks or neuromorphic hardware. For example, wake-sleep learning in networks characterized by Boltzmann distributions builds on the assumption of symmetric connectivity. Similarly, the error backpropagation algorithm is notoriously plagued by the weight transport problem between the representation and the error stream. Existing solutions such as feedback alignment tend to circumvent the problem by deferring to the robustness of these algorithms to weight asymmetry. However, they are known to scale poorly with network size and depth. We introduce spike-based alignment learning (SAL), a complementary learning rule for spiking neural networks, which uses spike timing statistics to extract and correct the asymmetry between effective reciprocal connections. Apart from being spike-based and fully local, our proposed mechanism takes advantage of noise. Based on an interplay between Hebbian and anti-Hebbian plasticity, synapses can thereby recover the true local gradient. This also alleviates discrepancies that arise from neuron and synapse variability -- an omnipresent property of physical neuronal networks. We demonstrate the efficacy of our mechanism using different spiking network models. First, we show how SAL can significantly improve convergence to the target distribution in probabilistic spiking networks as compared to Hebbian plasticity alone. Second, in neuronal hierarchies based on cortical microcircuits, we show how our proposed mechanism effectively enables the alignment of feedback weights to the forward pathway, thus allowing the backpropagation of correct feedback errors.
comment: 19 pages, 9 figures
☆ Weighted Euclidean Distance Matrices over Mixed Continuous and Categorical Inputs for Gaussian Process Models
Gaussian Process (GP) models are widely utilized as surrogate models in scientific and engineering fields. However, standard GP models are limited to continuous variables due to the difficulties in establishing correlation structures for categorical variables. To overcome this limitati on, we introduce WEighted Euclidean distance matrices Gaussian Process (WEGP). WEGP constructs the kernel function for each categorical input by estimating the Euclidean distance matrix (EDM) among all categorical choices of this input. The EDM is represented as a linear combination of several predefined base EDMs, each scaled by a positive weight. The weights, along with other kernel hyperparameters, are inferred using a fully Bayesian framework. We analyze the predictive performance of WEGP theoretically. Numerical experiments validate the accuracy of our GP model, and by WEGP, into Bayesian Optimization (BO), we achieve superior performance on both synthetic and real-world optimization problems.
☆ Leveraging Self-Supervised Learning Methods for Remote Screening of Subjects with Paroxysmal Atrial Fibrillation
The integration of Artificial Intelligence (AI) into clinical research has great potential to reveal patterns that are difficult for humans to detect, creating impactful connections between inputs and clinical outcomes. However, these methods often require large amounts of labeled data, which can be difficult to obtain in healthcare due to strict privacy laws and the need for experts to annotate data. This requirement creates a bottleneck when investigating unexplored clinical questions. This study explores the application of Self-Supervised Learning (SSL) as a way to obtain preliminary results from clinical studies with limited sized cohorts. To assess our approach, we focus on an underexplored clinical task: screening subjects for Paroxysmal Atrial Fibrillation (P-AF) using remote monitoring, single-lead ECG signals captured during normal sinus rhythm. We evaluate state-of-the-art SSL methods alongside supervised learning approaches, where SSL outperforms supervised learning in this task of interest. More importantly, it prevents misleading conclusions that may arise from poor performance in the latter paradigm when dealing with limited cohort settings.
comment: Under revision
☆ ZAPBench: A Benchmark for Whole-Brain Activity Prediction in Zebrafish
Data-driven benchmarks have led to significant progress in key scientific modeling domains including weather and structural biology. Here, we introduce the Zebrafish Activity Prediction Benchmark (ZAPBench) to measure progress on the problem of predicting cellular-resolution neural activity throughout an entire vertebrate brain. The benchmark is based on a novel dataset containing 4d light-sheet microscopy recordings of over 70,000 neurons in a larval zebrafish brain, along with motion stabilized and voxel-level cell segmentations of these data that facilitate development of a variety of forecasting methods. Initial results from a selection of time series and volumetric video modeling approaches achieve better performance than naive baseline methods, but also show room for further improvement. The specific brain used in the activity recording is also undergoing synaptic-level anatomical mapping, which will enable future integration of detailed structural information into forecasting methods.
☆ Smoothing the Shift: Towards Stable Test-Time Adaptation under Complex Multimodal Noises ICLR 2025
Test-Time Adaptation (TTA) aims to tackle distribution shifts using unlabeled test data without access to the source data. In the context of multimodal data, there are more complex noise patterns than unimodal data such as simultaneous corruptions for multiple modalities and missing modalities. Besides, in real-world applications, corruptions from different distribution shifts are always mixed. Existing TTA methods always fail in such multimodal scenario because the abrupt distribution shifts will destroy the prior knowledge from the source model, thus leading to performance degradation. To this end, we reveal a new challenge named multimodal wild TTA. To address this challenging problem, we propose two novel strategies: sample identification with interquartile range Smoothing and unimodal assistance, and Mutual information sharing (SuMi). SuMi smooths the adaptation process by interquartile range which avoids the abrupt distribution shifts. Then, SuMi fully utilizes the unimodal features to select low-entropy samples with rich multimodal information for optimization. Furthermore, mutual information sharing is introduced to align the information, reduce the discrepancies and enhance the information utilization across different modalities. Extensive experiments on two public datasets show the effectiveness and superiority over existing methods under the complex noise patterns in multimodal data. Code is available at https://github.com/zrguo/SuMi.
comment: Accepted at ICLR 2025
☆ Reinforcement Learning-based Threat Assessment
In some game scenarios, due to the uncertainty of the number of enemy units and the priority of various attributes, the evaluation of the threat level of enemy units as well as the screening has been a challenging research topic, and the core difficulty lies in how to reasonably set the priority of different attributes in order to achieve quantitative evaluation of the threat. In this paper, we innovatively transform the problem of threat assessment into a reinforcement learning problem, and through systematic reinforcement learning training, we successfully construct an efficient neural network evaluator. The evaluator can not only comprehensively integrate the multidimensional attribute features of the enemy, but also effectively combine our state information, thus realizing a more accurate and scientific threat assessment.
comment: 10 pages,9 figures
☆ Lightweight Channel-wise Dynamic Fusion Model: Non-stationary Time Series Forecasting via Entropy Analysis
Non-stationarity is an intrinsic property of real-world time series and plays a crucial role in time series forecasting. Previous studies primarily adopt instance normalization to attenuate the non-stationarity of original series for better predictability. However, instance normalization that directly removes the inherent non-stationarity can lead to three issues: (1) disrupting global temporal dependencies, (2) ignoring channel-specific differences, and (3) producing over-smoothed predictions. To address these issues, we theoretically demonstrate that variance can be a valid and interpretable proxy for quantifying non-stationarity of time series. Based on the analysis, we propose a novel lightweight \textit{C}hannel-wise \textit{D}ynamic \textit{F}usion \textit{M}odel (\textit{CDFM}), which selectively and dynamically recovers intrinsic non-stationarity of the original series, while keeping the predictability of normalized series. First, we design a Dual-Predictor Module, which involves two branches: a Time Stationary Predictor for capturing stable patterns and a Time Non-stationary Predictor for modeling global dynamics patterns. Second, we propose a Fusion Weight Learner to dynamically characterize the intrinsic non-stationary information across different samples based on variance. Finally, we introduce a Channel Selector to selectively recover non-stationary information from specific channels by evaluating their non-stationarity, similarity, and distribution consistency, enabling the model to capture relevant dynamic features and avoid overfitting. Comprehensive experiments on seven time series datasets demonstrate the superiority and generalization capabilities of CDFM.
☆ A generalized approach to label shift: the Conditional Probability Shift Model
In many practical applications of machine learning, a discrepancy often arises between a source distribution from which labeled training examples are drawn and a target distribution for which only unlabeled data is observed. Traditionally, two main scenarios have been considered to address this issue: covariate shift (CS), where only the marginal distribution of features changes, and label shift (LS), which involves a change in the class variable's prior distribution. However, these frameworks do not encompass all forms of distributional shift. This paper introduces a new setting, Conditional Probability Shift (CPS), which captures the case when the conditional distribution of the class variable given some specific features changes while the distribution of remaining features given the specific features and the class is preserved. For this scenario we present the Conditional Probability Shift Model (CPSM) based on modeling the class variable's conditional probabilities using multinomial regression. Since the class variable is not observed for the target data, the parameters of the multinomial model for its distribution are estimated using the Expectation-Maximization algorithm. The proposed method is generic and can be combined with any probabilistic classifier. The effectiveness of CPSM is demonstrated through experiments on synthetic datasets and a case study using the MIMIC medical database, revealing its superior balanced classification accuracy on the target data compared to existing methods, particularly in situations situations of conditional distribution shift and no apriori distribution shift, which are not detected by LS-based methods.
☆ To Vaccinate or not to Vaccinate? Analyzing $\mathbb{X}$ Power over the Pandemic
The COVID-19 pandemic has profoundly affected the normal course of life -- from lock-downs and virtual meetings to the unprecedentedly swift creation of vaccines. To halt the COVID-19 pandemic, the world has started preparing for the global vaccine roll-out. In an effort to navigate the immense volume of information about COVID-19, the public has turned to social networks. Among them, $\mathbb{X}$ (formerly Twitter) has played a key role in distributing related information. Most people are not trained to interpret medical research and remain skeptical about the efficacy of new vaccines. Measuring their reactions and perceptions is gaining significance in the fight against COVID-19. To assess the public perception regarding the COVID-19 vaccine, our work applies a sentiment analysis approach, using natural language processing of $\mathbb{X}$ data. We show how to use textual analytics and textual data visualization to discover early insights (for example, by analyzing the most frequently used keywords and hashtags). Furthermore, we look at how people's sentiments vary across the countries. Our results indicate that although the overall reaction to the vaccine is positive, there are also negative sentiments associated with the tweets, especially when examined at the country level. Additionally, from the extracted tweets, we manually labeled 100 tweets as positive and 100 tweets as negative and trained various One-Class Classifiers (OCCs). The experimental results indicate that the S-SVDD classifiers outperform other OCCs.
☆ Towards a robust R2D2 paradigm for radio-interferometric imaging: revisiting DNN training and architecture
The R2D2 Deep Neural Network (DNN) series was recently introduced for image formation in radio interferometry. It can be understood as a learned version of CLEAN, whose minor cycles are substituted with DNNs. We revisit R2D2 on the grounds of series convergence, training methodology, and DNN architecture, improving its robustness in terms of generalisability beyond training conditions, capability to deliver high data fidelity, and epistemic uncertainty. Firstly, while still focusing on telescope-specific training, we enhance the learning process by randomising Fourier sampling integration times, incorporating multi-scan multi-noise configurations, and varying imaging settings, including pixel resolution and visibility-weighting scheme. Secondly, we introduce a convergence criterion whereby the reconstruction process stops when the data residual is compatible with noise, rather than simply using all available DNNs. This not only increases the reconstruction efficiency by reducing its computational cost, but also refines training by pruning out the data/image pairs for which optimal data fidelity is reached before training the next DNN. Thirdly, we substitute R2D2's early U-Net DNN with a novel architecture (U-WDSR) combining U-Net and WDSR, which leverages wide activation, dense connections, weight normalisation, and low-rank convolution to improve feature reuse and reconstruction precision. As previously, R2D2 was trained for monochromatic intensity imaging with the Very Large Array (VLA) at fixed $512 \times 512$ image size. Simulations on a wide range of inverse problems and a case study on real data reveal that the new R2D2 model consistently outperforms its earlier version in image reconstruction quality, data fidelity, and epistemic uncertainty.
comment: 17 pages, 6 figures
☆ Disentangled Knowledge Tracing for Alleviating Cognitive Bias
In the realm of Intelligent Tutoring System (ITS), the accurate assessment of students' knowledge states through Knowledge Tracing (KT) is crucial for personalized learning. However, due to data bias, $\textit{i.e.}$, the unbalanced distribution of question groups ($\textit{e.g.}$, concepts), conventional KT models are plagued by cognitive bias, which tends to result in cognitive underload for overperformers and cognitive overload for underperformers. More seriously, this bias is amplified with the exercise recommendations by ITS. After delving into the causal relations in the KT models, we identify the main cause as the confounder effect of students' historical correct rate distribution over question groups on the student representation and prediction score. Towards this end, we propose a Disentangled Knowledge Tracing (DisKT) model, which separately models students' familiar and unfamiliar abilities based on causal effects and eliminates the impact of the confounder in student representation within the model. Additionally, to shield the contradictory psychology ($\textit{e.g.}$, guessing and mistaking) in the students' biased data, DisKT introduces a contradiction attention mechanism. Furthermore, DisKT enhances the interpretability of the model predictions by integrating a variant of Item Response Theory. Experimental results on 11 benchmarks and 3 synthesized datasets with different bias strengths demonstrate that DisKT significantly alleviates cognitive bias and outperforms 16 baselines in evaluation accuracy.
☆ SAGE-Amine: Generative Amine Design with Multi-Property Optimization for Efficient CO2 Capture
Efficient CO2 capture is vital for mitigating climate change, with amine-based solvents being widely used due to their strong reactivity with CO2. However, optimizing key properties such as basicity, viscosity, and absorption capacity remains challenging, as traditional methods rely on labor-intensive experimentation and predefined chemical databases, limiting the exploration of novel solutions. Here, SAGE-Amine was introduced, a generative modeling approach that integrates Scoring-Assisted Generative Exploration (SAGE) with quantitative structure-property relationship models to design new amines tailored for CO2 capture. Unlike conventional virtual screening restricted to existing compounds, SAGE-Amine generates novel amines by leveraging autoregressive natural language processing models trained on amine datasets. SAGE-Amine identified known amines for CO2 capture from scratch and successfully performed single-property optimization, increasing basicity or reducing viscosity or vapor pressure. Furthermore, it facilitated multi-property optimization, simultaneously achieving high basicity with low viscosity and vapor pressure. The 10 top-ranked amines were suggested using SAGE-Amine and their thermodynamic properties were further assessed using COSMO-RS simulations, confirming their potential for CO2 capture. These results highlight the potential of generative modeling in accelerating the discovery of amine solvents and expanding the possibilities for industrial CO2 capture applications.
comment: 33 pages, 5 figures
☆ A Theory of Initialisation's Impact on Specialisation
Prior work has demonstrated a consistent tendency in neural networks engaged in continual learning tasks, wherein intermediate task similarity results in the highest levels of catastrophic interference. This phenomenon is attributed to the network's tendency to reuse learned features across tasks. However, this explanation heavily relies on the premise that neuron specialisation occurs, i.e. the emergence of localised representations. Our investigation challenges the validity of this assumption. Using theoretical frameworks for the analysis of neural networks, we show a strong dependence of specialisation on the initial condition. More precisely, we show that weight imbalance and high weight entropy can favour specialised solutions. We then apply these insights in the context of continual learning, first showing the emergence of a monotonic relation between task-similarity and forgetting in non-specialised networks. {Finally, we show that specialization by weight imbalance is beneficial on the commonly employed elastic weight consolidation regularisation technique.
comment: 10 pages, 7 figures
☆ LTL Verification of Memoryful Neural Agents AAMAS 2025
We present a framework for verifying Memoryful Neural Multi-Agent Systems (MN-MAS) against full Linear Temporal Logic (LTL) specifications. In MN-MAS, agents interact with a non-deterministic, partially observable environment. Examples of MN-MAS include multi-agent systems based on feed-forward and recurrent neural networks or state-space models. Different from previous approaches, we support the verification of both bounded and unbounded LTL specifications. We leverage well-established bounded model checking techniques, including lasso search and invariant synthesis, to reduce the verification problem to that of constraint solving. To solve these constraints, we develop efficient methods based on bound propagation, mixed-integer linear programming, and adaptive splitting. We evaluate the effectiveness of our algorithms in single and multi-agent environments from the Gymnasium and PettingZoo libraries, verifying unbounded specifications for the first time and improving the verification time for bounded specifications by an order of magnitude compared to the SoA.
comment: 11 pages, 2 figures, accepted at AAMAS 2025 conference
☆ ROCKET-2: Steering Visuomotor Policy via Cross-View Goal Alignment
We aim to develop a goal specification method that is semantically clear, spatially sensitive, and intuitive for human users to guide agent interactions in embodied environments. Specifically, we propose a novel cross-view goal alignment framework that allows users to specify target objects using segmentation masks from their own camera views rather than the agent's observations. We highlight that behavior cloning alone fails to align the agent's behavior with human intent when the human and agent camera views differ significantly. To address this, we introduce two auxiliary objectives: cross-view consistency loss and target visibility loss, which explicitly enhance the agent's spatial reasoning ability. According to this, we develop ROCKET-2, a state-of-the-art agent trained in Minecraft, achieving an improvement in the efficiency of inference 3x to 6x. We show ROCKET-2 can directly interpret goals from human camera views for the first time, paving the way for better human-agent interaction.
☆ A Systematic Literature Review on Safety of the Intended Functionality for Automated Driving Systems
In the automobile industry, ensuring the safety of automated vehicles equipped with the Automated Driving System (ADS) is becoming a significant focus due to the increasing development and deployment of automated driving. Automated driving depends on sensing both the external and internal environments of a vehicle, utilizing perception sensors and algorithms, and Electrical/Electronic (E/E) systems for situational awareness and response. ISO 21448 is the standard for Safety of the Intended Functionality (SOTIF) that aims to ensure that the ADS operate safely within their intended functionality. SOTIF focuses on preventing or mitigating potential hazards that may arise from the limitations or failures of the ADS, including hazards due to insufficiencies of specification, or performance insufficiencies, as well as foreseeable misuse of the intended functionality. However, the challenge lies in ensuring the safety of vehicles despite the limited availability of extensive and systematic literature on SOTIF. To address this challenge, a Systematic Literature Review (SLR) on SOTIF for the ADS is performed following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The objective is to methodically gather and analyze the existing literature on SOTIF. The major contributions of this paper are: (i) presenting a summary of the literature by synthesizing and organizing the collective findings, methodologies, and insights into distinct thematic groups, and (ii) summarizing and categorizing the acknowledged limitations based on data extracted from an SLR of 51 research papers published between 2018 and 2023. Furthermore, research gaps are determined, and future research directions are proposed.
comment: Scheduled to be published in SAE journal as technical paper as a part of non-technical event and will be available as open access in 2025
☆ Union of Experts: Adapting Hierarchical Routing to Equivalently Decomposed Transformer
Mixture-of-Experts (MoE) enhances model performance while maintaining computational efficiency, making it well-suited for large-scale applications. However, expert in exist MoE paradigm works as an individual, thereby lacking high-quality expert interactions. Moreover, they have not been effectively extended to attention block, which constrains further efficiency improvements. To tackle these issues, we propose Union-of-Experts (UoE), which decomposes transformer into an equitant group of experts, and then implement dynamic routing on input data and experts. Our approach advances MoE design with three key innovations: (1) We conducted equitant expert decomposition on both MLP blocks and attention blocks based on matrix partition in tensor parallelism. (2) We developed two routing paradigms: patch wise data selection and expert selection, to apply routing across different levels. (3) We design the architecture of UoE model, including Selective Multi-Head Attention (SMHA) and Union-of-MLP-Experts (UoME). (4) We develop parallel implementation of UoE's routing and computation operation, and optimize efficiency based on the hardware processing analysis. The experiments demonstrate that the model employed with UoE surpass Full Attention, state-of-art MoEs and efficient transformers in several tasks across image and natural language domains. The source codes are available at https://github.com/YujiaoYang-work/UoE.
comment: 17 pages, 6 figures, 5 tables
☆ The Distributionally Robust Optimization Model of Sparse Principal Component Analysis
We consider sparse principal component analysis (PCA) under a stochastic setting where the underlying probability distribution of the random parameter is uncertain. This problem is formulated as a distributionally robust optimization (DRO) model based on a constructive approach to capturing uncertainty in the covariance matrix, which constitutes a nonsmooth constrained min-max optimization problem. We further prove that the inner maximization problem admits a closed-form solution, reformulating the original DRO model into an equivalent minimization problem on the Stiefel manifold. This transformation leads to a Riemannian optimization problem with intricate nonsmooth terms, a challenging formulation beyond the reach of existing algorithms. To address this issue, we devise an efficient smoothing manifold proximal gradient algorithm. We prove the Riemannian gradient consistency and global convergence of our algorithm to a stationary point of the nonsmooth minimization problem. Moreover, we establish the iteration complexity of our algorithm. Finally, numerical experiments are conducted to validate the effectiveness and scalability of our algorithm, as well as to highlight the necessity and rationality of adopting the DRO model for sparse PCA.
☆ Joint Out-of-Distribution Filtering and Data Discovery Active Learning
As the data demand for deep learning models increases, active learning (AL) becomes essential to strategically select samples for labeling, which maximizes data efficiency and reduces training costs. Real-world scenarios necessitate the consideration of incomplete data knowledge within AL. Prior works address handling out-of-distribution (OOD) data, while another research direction has focused on category discovery. However, a combined analysis of real-world considerations combining AL with out-of-distribution data and category discovery remains unexplored. To address this gap, we propose Joint Out-of-distribution filtering and data Discovery Active learning (Joda) , to uniquely address both challenges simultaneously by filtering out OOD data before selecting candidates for labeling. In contrast to previous methods, we deeply entangle the training procedure with filter and selection to construct a common feature space that aligns known and novel categories while separating OOD samples. Unlike previous works, Joda is highly efficient and completely omits auxiliary models and training access to the unlabeled pool for filtering or selection. In extensive experiments on 18 configurations and 3 metrics, \ours{} consistently achieves the highest accuracy with the best class discovery to OOD filtering balance compared to state-of-the-art competitor approaches.
☆ Joint Tensor and Inter-View Low-Rank Recovery for Incomplete Multiview Clustering
Incomplete multiview clustering (IMVC) has gained significant attention for its effectiveness in handling missing sample challenges across various views in real-world multiview clustering applications. Most IMVC approaches tackle this problem by either learning consensus representations from available views or reconstructing missing samples using the underlying manifold structure. However, the reconstruction of learned similarity graph tensor in prior studies only exploits the low-tubal-rank information, neglecting the exploration of inter-view correlations. This paper propose a novel joint tensor and inter-view low-rank Recovery (JTIV-LRR), framing IMVC as a joint optimization problem that integrates incomplete similarity graph learning and tensor representation recovery. By leveraging both intra-view and inter-view low rank information, the method achieves robust estimation of the complete similarity graph tensor through sparse noise removal and low-tubal-rank constraints along different modes. Extensive experiments on both synthetic and real-world datasets demonstrate the superiority of the proposed approach, achieving significant improvements in clustering accuracy and robustness compared to state-of-the-art methods.
comment: The paper is under review at IEEE Transactions on Knowledge and Data Engineering
☆ NodeNAS: Node-Specific Graph Neural Architecture Search for Out-of-Distribution Generalization DASFAA2025
Graph neural architecture search (GraphNAS) has demonstrated advantages in mitigating performance degradation of graph neural networks (GNNs) due to distribution shifts. Recent approaches introduce weight sharing across tailored architectures, generating unique GNN architectures for each graph end-to-end. However, existing GraphNAS methods do not account for distribution patterns across different graphs and heavily rely on extensive training data. With sparse or single training graphs, these methods struggle to discover optimal mappings between graphs and architectures, failing to generalize to out-of-distribution (OOD) data. In this paper, we propose node-specific graph neural architecture search(NodeNAS), which aims to tailor distinct aggregation methods for different nodes through disentangling node topology and graph distribution with limited datasets. We further propose adaptive aggregation attention based multi-dim NodeNAS method(MNNAS), which learns an node-specific architecture customizer with good generalizability. Specifically, we extend the vertical depth of the search space, supporting simultaneous node-specific architecture customization across multiple dimensions. Moreover, we model the power-law distribution of node degrees under varying assortativity, encoding structure invariant information to guide architecture customization across each dimension. Extensive experiments across supervised and unsupervised tasks demonstrate that MNNAS surpasses state-of-the-art algorithms and achieves excellent OOD generalization.
comment: Accepted by DASFAA2025
☆ BRIDGE: Bootstrapping Text to Control Time-Series Generation via Multi-Agent Iterative Optimization and Diffusion Modelling
Time-series Generation (TSG) is a prominent research area with broad applications in simulations, data augmentation, and counterfactual analysis. While existing methods have shown promise in unconditional single-domain TSG, real-world applications demand for cross-domain approaches capable of controlled generation tailored to domain-specific constraints and instance-level requirements. In this paper, we argue that text can provide semantic insights, domain information and instance-specific temporal patterns, to guide and improve TSG. We introduce ``Text-Controlled TSG'', a task focused on generating realistic time series by incorporating textual descriptions. To address data scarcity in this setting, we propose a novel LLM-based Multi-Agent framework that synthesizes diverse, realistic text-to-TS datasets. Furthermore, we introduce BRIDGE, a hybrid text-controlled TSG framework that integrates semantic prototypes with text description for supporting domain-level guidance. This approach achieves state-of-the-art generation fidelity on 11 of 12 datasets, and improves controllability by 12.52% on MSE and 6.34% MAE compared to no text input generation, highlighting its potential for generating tailored time-series data.
comment: Preprint. Work in progress
☆ Artificial Intelligence in Reactor Physics: Current Status and Future Prospects
Reactor physics is the study of neutron properties, focusing on using models to examine the interactions between neutrons and materials in nuclear reactors. Artificial intelligence (AI) has made significant contributions to reactor physics, e.g., in operational simulations, safety design, real-time monitoring, core management and maintenance. This paper presents a comprehensive review of AI approaches in reactor physics, especially considering the category of Machine Learning (ML), with the aim of describing the application scenarios, frontier topics, unsolved challenges and future research directions. From equation solving and state parameter prediction to nuclear industry applications, this paper provides a step-by-step overview of ML methods applied to steady-state, transient and combustion problems. Most literature works achieve industry-demanded models by enhancing the efficiency of deterministic methods or correcting uncertainty methods, which leads to successful applications. However, research on ML methods in reactor physics is somewhat fragmented, and the ability to generalize models needs to be strengthened. Progress is still possible, especially in addressing theoretical challenges and enhancing industrial applications such as building surrogate models and digital twins.
comment: 33 pages, 6 figures
☆ Decentralized Reinforcement Learning for Multi-Agent Multi-Resource Allocation via Dynamic Cluster Agreements
This paper addresses the challenge of allocating heterogeneous resources among multiple agents in a decentralized manner. Our proposed method, LGTC-IPPO, builds upon Independent Proximal Policy Optimization (IPPO) by integrating dynamic cluster consensus, a mechanism that allows agents to form and adapt local sub-teams based on resource demands. This decentralized coordination strategy reduces reliance on global information and enhances scalability. We evaluate LGTC-IPPO against standard multi-agent reinforcement learning baselines and a centralized expert solution across a range of team sizes and resource distributions. Experimental results demonstrate that LGTC-IPPO achieves more stable rewards, better coordination, and robust performance even as the number of agents or resource types increases. Additionally, we illustrate how dynamic clustering enables agents to reallocate resources efficiently also for scenarios with discharging resources.
☆ Tight Gap-Dependent Memory-Regret Trade-Off for Single-Pass Streaming Stochastic Multi-Armed Bandits
We study the problem of minimizing gap-dependent regret for single-pass streaming stochastic multi-armed bandits (MAB). In this problem, the $n$ arms are present in a stream, and at most $m 0}\Delta_i^{1 - 2\alpha}\Big)$ and another applicable for $m<\frac{2}{3}n$ with regret at most $O_\alpha\Big(\frac{T^{\frac{1}{\alpha+1}}}{m^{\frac{1}{\alpha+1}}}\displaystyle\sum_{i:\Delta_i > 0}\Delta_i^{1 - 2\alpha}\Big)$. We also prove matching lower bounds for both cases by showing that for any constant $\alpha\ge 1$ and any $m\leq k < n$, there exists a set of hard instances on which the regret of any algorithm is $\Omega_\alpha\Big(\frac{(k-m+1) T^{\frac{1}{\alpha+1}}}{k^{1 + \frac{1}{\alpha+1}}} \sum_{i:\Delta_i > 0}\Delta_i^{1-2\alpha}\Big)$. This is the first tight gap-dependent regret bound for streaming MAB. Prior to our work, an $O\Big(\sum_{i\colon\Delta>0} \frac{\sqrt{T}\log T}{\Delta_i}\Big)$ upper bound for the special case of $\alpha=1$ and $m=O(1)$ was established by Agarwal, Khanna and Patil (COLT'22). In contrast, our results provide the correct order of regret as $\Theta\Big(\frac{1}{\sqrt{m}}\sum_{i\colon\Delta>0}\frac{\sqrt{T}}{\Delta_i}\Big)$.
☆ Aggregation Strategies for Efficient Annotation of Bioacoustic Sound Events Using Active Learning
The vast amounts of audio data collected in Sound Event Detection (SED) applications require efficient annotation strategies to enable supervised learning. Manual labeling is expensive and time-consuming, making Active Learning (AL) a promising approach for reducing annotation effort. We introduce Top K Entropy, a novel uncertainty aggregation strategy for AL that prioritizes the most uncertain segments within an audio recording, instead of averaging uncertainty across all segments. This approach enables the selection of entire recordings for annotation, improving efficiency in sparse data scenarios. We compare Top K Entropy to random sampling and Mean Entropy, and show that fewer labels can lead to the same model performance, particularly in datasets with sparse sound events. Evaluations are conducted on audio mixtures of sound recordings from parks with meerkat, dog, and baby crying sound events, representing real-world bioacoustic monitoring scenarios. Using Top K Entropy for active learning, we can achieve comparable performance to training on the fully labeled dataset with only 8% of the labels. Top K Entropy outperforms Mean Entropy, suggesting that it is best to let the most uncertain segments represent the uncertainty of an audio file. The findings highlight the potential of AL for scalable annotation in audio and time-series applications, including bioacoustics.
☆ A Transformer-Based Framework for Greek Sign Language Production using Extended Skeletal Motion Representations
Sign Languages are the primary form of communication for Deaf communities across the world. To break the communication barriers between the Deaf and Hard-of-Hearing and the hearing communities, it is imperative to build systems capable of translating the spoken language into sign language and vice versa. Building on insights from previous research, we propose a deep learning model for Sign Language Production (SLP), which to our knowledge is the first attempt on Greek SLP. We tackle this task by utilizing a transformer-based architecture that enables the translation from text input to human pose keypoints, and the opposite. We evaluate the effectiveness of the proposed pipeline on the Greek SL dataset Elementary23, through a series of comparative analyses and ablation studies. Our pipeline's components, which include data-driven gloss generation, training through video to text translation and a scheduling algorithm for teacher forcing - auto-regressive decoding seem to actively enhance the quality of produced SL videos.
☆ InfoGNN: End-to-end deep learning on mesh via graph neural networks
3D models are widely used in various industries, and mesh data has become an indispensable part of 3D modeling because of its unique advantages. Mesh data can provide an intuitive and practical expression of rich 3D information. However, its disordered, irregular data structure and complex surface information make it challenging to apply with deep learning models directly. Traditional mesh data processing methods often rely on mesh models with many limitations, such as manifold, which restrict their application scopes in reality and do not fully utilize the advantages of mesh models. This paper proposes a novel end-to-end framework for addressing the challenges associated with deep learning in mesh models centered around graph neural networks (GNN) and is titled InfoGNN. InfoGNN treats the mesh model as a graph, which enables it to handle irregular mesh data efficiently. Moreover, we propose InfoConv and InfoMP modules, which utilize the position information of the points and fully use the static information such as face normals, dihedral angles, and dynamic global feature information to fully use all kinds of data. In addition, InfoGNN is an end-to-end framework, and we simplify the network design to make it more efficient, paving the way for efficient deep learning of complex 3D models. We conducted experiments on several publicly available datasets, and the results show that InfoGNN achieves excellent performance in mesh classification and segmentation tasks.
☆ Wyckoff Transformer: Generation of Symmetric Crystals
Symmetry rules that atoms obey when they bond together to form an ordered crystal play a fundamental role in determining their physical, chemical, and electronic properties such as electrical and thermal conductivity, optical and polarization behavior, and mechanical strength. Almost all known crystalline materials have internal symmetry. Consistently generating stable crystal structures is still an open challenge, specifically because such symmetry rules are not accounted for. To address this issue, we propose WyFormer, a generative model for materials conditioned on space group symmetry. We use Wyckoff positions as the basis for an elegant, compressed, and discrete structure representation. To model the distribution, we develop a permutation-invariant autoregressive model based on the Transformer and an absence of positional encoding. WyFormer has a unique and powerful synergy of attributes, proven by extensive experimentation: best-in-class symmetry-conditioned generation, physics-motivated inductive bias, competitive stability of the generated structures, competitive material property prediction quality, and unparalleled inference speed.
comment: https://github.com/SymmetryAdvantage/WyckoffTransformer
☆ A Binary Classification Social Network Dataset for Graph Machine Learning
Social networks have a vast range of applications with graphs. The available benchmark datasets are citation, co-occurrence, e-commerce networks, etc, with classes ranging from 3 to 15. However, there is no benchmark classification social network dataset for graph machine learning. This paper fills the gap and presents the Binary Classification Social Network Dataset (\textit{BiSND}), designed for graph machine learning applications to predict binary classes. We present the BiSND in \textit{tabular and graph} formats to verify its robustness across classical and advanced machine learning. We employ a diverse set of classifiers, including four traditional machine learning algorithms (Decision Trees, K-Nearest Neighbour, Random Forest, XGBoost), one Deep Neural Network (multi-layer perceptrons), one Graph Neural Network (Graph Convolutional Network), and three state-of-the-art Graph Contrastive Learning methods (BGRL, GRACE, DAENS). Our findings reveal that BiSND is suitable for classification tasks, with F1-scores ranging from 67.66 to 70.15, indicating promising avenues for future enhancements.
☆ Robust detection of overlapping bioacoustic sound events
We propose a method for accurately detecting bioacoustic sound events that is robust to overlapping events, a common issue in domains such as ethology, ecology and conservation. While standard methods employ a frame-based, multi-label approach, we introduce an onset-based detection method which we name Voxaboxen. It takes inspiration from object detection methods in computer vision, but simultaneously takes advantage of recent advances in self-supervised audio encoders. For each time window, Voxaboxen predicts whether it contains the start of a vocalization and how long the vocalization is. It also does the same in reverse, predicting whether each window contains the end of a vocalization, and how long ago it started. The two resulting sets of bounding boxes are then fused using a graph-matching algorithm. We also release a new dataset designed to measure performance on detecting overlapping vocalizations. This consists of recordings of zebra finches annotated with temporally-strong labels and showing frequent overlaps. We test Voxaboxen on seven existing data sets and on our new data set. We compare Voxaboxen to natural baselines and existing sound event detection methods and demonstrate SotA results. Further experiments show that improvements are robust to frequent vocalization overlap.
☆ An Accelerated Alternating Partial Bregman Algorithm for ReLU-based Matrix Decomposition
Despite the remarkable success of low-rank estimation in data mining, its effectiveness diminishes when applied to data that inherently lacks low-rank structure. To address this limitation, in this paper, we focus on non-negative sparse matrices and aim to investigate the intrinsic low-rank characteristics of the rectified linear unit (ReLU) activation function. We first propose a novel nonlinear matrix decomposition framework incorporating a comprehensive regularization term designed to simultaneously promote useful structures in clustering and compression tasks, such as low-rankness, sparsity, and non-negativity in the resulting factors. This formulation presents significant computational challenges due to its multi-block structure, non-convexity, non-smoothness, and the absence of global gradient Lipschitz continuity. To address these challenges, we develop an accelerated alternating partial Bregman proximal gradient method (AAPB), whose distinctive feature lies in its capability to enable simultaneous updates of multiple variables. Under mild and theoretically justified assumptions, we establish both sublinear and global convergence properties of the proposed algorithm. Through careful selection of kernel generating distances tailored to various regularization terms, we derive corresponding closed-form solutions while maintaining the $L$-smooth adaptable property always holds for any $L\ge 1$. Numerical experiments, on graph regularized clustering and sparse NMF basis compression confirm the effectiveness of our model and algorithm.
☆ Truthfulness of Decision-Theoretic Calibration Measures
Calibration measures quantify how much a forecaster's predictions violates calibration, which requires that forecasts are unbiased conditioning on the forecasted probabilities. Two important desiderata for a calibration measure are its decision-theoretic implications (i.e., downstream decision-makers that best-respond to the forecasts are always no-regret) and its truthfulness (i.e., a forecaster approximately minimizes error by always reporting the true probabilities). Existing measures satisfy at most one of the properties, but not both. We introduce a new calibration measure termed subsampled step calibration, $\mathsf{StepCE}^{\textsf{sub}}$, that is both decision-theoretic and truthful. In particular, on any product distribution, $\mathsf{StepCE}^{\textsf{sub}}$ is truthful up to an $O(1)$ factor whereas prior decision-theoretic calibration measures suffer from an $e^{-\Omega(T)}$-$\Omega(\sqrt{T})$ truthfulness gap. Moreover, in any smoothed setting where the conditional probability of each event is perturbed by a noise of magnitude $c > 0$, $\mathsf{StepCE}^{\textsf{sub}}$ is truthful up to an $O(\sqrt{\log(1/c)})$ factor, while prior decision-theoretic measures have an $e^{-\Omega(T)}$-$\Omega(T^{1/3})$ truthfulness gap. We also prove a general impossibility result for truthful decision-theoretic forecasting: any complete and decision-theoretic calibration measure must be discontinuous and non-truthful in the non-smoothed setting.
☆ Teaching Metric Distance to Autoregressive Multimodal Foundational Models
As large language models expand beyond natural language to domains such as mathematics, multimodal understanding, and embodied agents, tokens increasingly reflect metric relationships rather than purely linguistic meaning. We introduce DIST2Loss, a distance-aware framework designed to train autoregressive discrete models by leveraging predefined distance relationships among output tokens. At its core, DIST2Loss transforms continuous exponential family distributions derived from inherent distance metrics into discrete, categorical optimization targets compatible with the models' architectures. This approach enables the models to learn and preserve meaningful distance relationships during token generation while maintaining compatibility with existing architectures. Empirical evaluations show consistent performance gains in diverse multimodal applications, including visual grounding, robotic manipulation, generative reward modeling, and image generation using vector-quantized features. These improvements are pronounced in cases of limited training data, highlighting DIST2Loss's effectiveness in resource-constrained settings.
☆ Iterative Value Function Optimization for Guided Decoding
While Reinforcement Learning from Human Feedback (RLHF) has become the predominant method for controlling language model outputs, it suffers from high computational costs and training instability. Guided decoding, especially value-guided methods, offers a cost-effective alternative by controlling outputs without re-training models. However, the accuracy of the value function is crucial for value-guided decoding, as inaccuracies can lead to suboptimal decision-making and degraded performance. Existing methods struggle with accurately estimating the optimal value function, leading to less effective control. We propose Iterative Value Function Optimization, a novel framework that addresses these limitations through two key components: Monte Carlo Value Estimation, which reduces estimation variance by exploring diverse trajectories, and Iterative On-Policy Optimization, which progressively improves value estimation through collecting trajectories from value-guided policies. Extensive experiments on text summarization, multi-turn dialogue, and instruction following demonstrate the effectiveness of value-guided decoding approaches in aligning language models. These approaches not only achieve alignment but also significantly reduce computational costs by leveraging principled value function optimization for efficient and effective control.
comment: 20 pages, 10 figures
☆ BdSLW401: Transformer-Based Word-Level Bangla Sign Language Recognition Using Relative Quantization Encoding (RQE)
Sign language recognition (SLR) for low-resource languages like Bangla suffers from signer variability, viewpoint variations, and limited annotated datasets. In this paper, we present BdSLW401, a large-scale, multi-view, word-level Bangla Sign Language (BdSL) dataset with 401 signs and 102,176 video samples from 18 signers in front and lateral views. To improve transformer-based SLR, we introduce Relative Quantization Encoding (RQE), a structured embedding approach anchoring landmarks to physiological reference points and quantize motion trajectories. RQE improves attention allocation by decreasing spatial variability, resulting in 44.3% WER reduction in WLASL100, 21.0% in SignBD-200, and significant gains in BdSLW60 and SignBD-90. However, fixed quantization becomes insufficient on large-scale datasets (e.g., WLASL2000), indicating the need for adaptive encoding strategies. Further, RQE-SF, an extended variant that stabilizes shoulder landmarks, achieves improvements in pose consistency at the cost of small trade-offs in lateral view recognition. The attention graphs prove that RQE improves model interpretability by focusing on the major articulatory features (fingers, wrists) and the more distinctive frames instead of global pose changes. Introducing BdSLW401 and demonstrating the effectiveness of RQE-enhanced structured embeddings, this work advances transformer-based SLR for low-resource languages and sets a benchmark for future research in this area.
☆ Controllable Motion Generation via Diffusion Modal Coupling
Diffusion models have recently gained significant attention in robotics due to their ability to generate multi-modal distributions of system states and behaviors. However, a key challenge remains: ensuring precise control over the generated outcomes without compromising realism. This is crucial for applications such as motion planning or trajectory forecasting, where adherence to physical constraints and task-specific objectives is essential. We propose a novel framework that enhances controllability in diffusion models by leveraging multi-modal prior distributions and enforcing strong modal coupling. This allows us to initiate the denoising process directly from distinct prior modes that correspond to different possible system behaviors, ensuring sampling to align with the training distribution. We evaluate our approach on motion prediction using the Waymo dataset and multi-task control in Maze2D environments. Experimental results show that our framework outperforms both guidance-based techniques and conditioned models with unimodal priors, achieving superior fidelity, diversity, and controllability, even in the absence of explicit conditioning. Overall, our approach provides a more reliable and scalable solution for controllable motion generation in robotics.
☆ Confidence HNC: A Network Flow Technique for Binary Classification with Noisy Labels
We consider here a classification method that balances two objectives: large similarity within the samples in the cluster, and large dissimilarity between the cluster and its complement. The method, referred to as HNC or SNC, requires seed nodes, or labeled samples, at least one of which is in the cluster and at least one in the complement. Other than that, the method relies only on the relationship between the samples. The contribution here is the new method in the presence of noisy labels, based on HNC, called Confidence HNC, in which we introduce confidence weights that allow the given labels of labeled samples to be violated, with a penalty that reflects the perceived correctness of each given label. If a label is violated then it is interpreted that the label was noisy. The method involves a representation of the problem as a graph problem with hyperparameters that is solved very efficiently by the network flow technique of parametric cut. We compare the performance of the new method with leading algorithms on both real and synthetic data with noisy labels and demonstrate that it delivers improved performance in terms of classification accuracy as well as noise detection capability.
☆ CQ CNN: A Hybrid Classical Quantum Convolutional Neural Network for Alzheimer's Disease Detection Using Diffusion Generated and U Net Segmented 3D MRI
The detection of Alzheimer disease (AD) from clinical MRI data is an active area of research in medical imaging. Recent advances in quantum computing, particularly the integration of parameterized quantum circuits (PQCs) with classical machine learning architectures, offer new opportunities to develop models that may outperform traditional methods. However, quantum machine learning (QML) remains in its early stages and requires further experimental analysis to better understand its behavior and limitations. In this paper, we propose an end to end hybrid classical quantum convolutional neural network (CQ CNN) for AD detection using clinically formatted 3D MRI data. Our approach involves developing a framework to make 3D MRI data usable for machine learning, designing and training a brain tissue segmentation model (Skull Net), and training a diffusion model to generate synthetic images for the minority class. Our converged models exhibit potential quantum advantages, achieving higher accuracy in fewer epochs than classical models. The proposed beta8 3 qubit model achieves an accuracy of 97.50%, surpassing state of the art (SOTA) models while requiring significantly fewer computational resources. In particular, the architecture employs only 13K parameters (0.48 MB), reducing the parameter count by more than 99.99% compared to current SOTA models. Furthermore, the diffusion-generated data used to train our quantum models, in conjunction with real samples, preserve clinical structural standards, representing a notable first in the field of QML. We conclude that CQCNN architecture like models, with further improvements in gradient optimization techniques, could become a viable option and even a potential alternative to classical models for AD detection, especially in data limited and resource constrained clinical settings.
comment: Application of hybrid quantum-classical machine learning for (early stage) disease detection
☆ DeLTa: A Decoding Strategy based on Logit Trajectory Prediction Improves Factuality and Reasoning Ability
Large Language Models (LLMs) are increasingly being used in real-world applications. However, concerns about the reliability of the content they generate persist, as it frequently deviates from factual correctness or exhibits deficiencies in logical reasoning. This paper proposes a novel decoding strategy aimed at enhancing both factual accuracy and inferential reasoning without requiring any modifications to the architecture or pre-trained parameters of LLMs. Our approach adjusts next-token probabilities by analyzing the trajectory of logits from lower to higher layers in Transformers and applying linear regression. We find that this Decoding by Logit Trajectory-based approach (DeLTa) effectively reinforces factuality and reasoning while mitigating incorrect generation. Experiments on TruthfulQA demonstrate that DeLTa attains up to a 4.9% improvement over the baseline. Furthermore, it enhances performance by up to 8.1% on StrategyQA and 7.3% on GSM8K, both of which demand strong reasoning capabilities.
comment: Source code is available at https://github.com/githubhyz/DeLTa
☆ GRADEO: Towards Human-Like Evaluation for Text-to-Video Generation via Multi-Step Reasoning
Recent great advances in video generation models have demonstrated their potential to produce high-quality videos, bringing challenges to effective evaluation. Unlike human evaluation, existing automated evaluation metrics lack high-level semantic understanding and reasoning capabilities for video, thus making them infeasible and unexplainable. To fill this gap, we curate GRADEO-Instruct, a multi-dimensional T2V evaluation instruction tuning dataset, including 3.3k videos from over 10 existing video generation models and multi-step reasoning assessments converted by 16k human annotations. We then introduce GRADEO, one of the first specifically designed video evaluation models, which grades AI-generated videos for explainable scores and assessments through multi-step reasoning. Experiments show that our method aligns better with human evaluations than existing methods. Furthermore, our benchmarking reveals that current video generation models struggle to produce content that aligns with human reasoning and complex real-world scenarios. The models, datasets, and codes will be released soon.
♻ ☆ Variational Best-of-N Alignment ICLR 2025
Best-of-N (BoN) is a popular and effective algorithm for aligning language models to human preferences. The algorithm works as follows: at inference time, N samples are drawn from the language model, and the sample with the highest reward, as judged by a reward model, is returned as the output. Despite its effectiveness, BoN is computationally expensive; it reduces sampling throughput by a factor of N. To make BoN more efficient at inference time, one strategy is to fine-tune the language model to mimic what BoN does during inference. To achieve this, we derive the distribution induced by the BoN algorithm. We then propose to fine-tune the language model to minimize backward KL divergence to the BoN distribution. Our approach is analogous to mean-field variational inference and, thus, we term it variational BoN (vBoN). To the extent this fine-tuning is successful and we end up with a good approximation, we have reduced the inference cost by a factor of N. Our experiments on controlled generation and summarization tasks show that BoN is the most effective alignment method, and our variational approximation to BoN achieves the closest performance to BoN and surpasses models fine-tuned using the standard KL-constrained RL objective. In the controlled generation task, vBoN appears more frequently on the Pareto frontier of reward and KL divergence compared to other alignment methods. In the summarization task, vBoN achieves high reward values across various sampling temperatures.
comment: Accepted at ICLR 2025
♻ ☆ UMGAD: Unsupervised Multiplex Graph Anomaly Detection
Graph anomaly detection (GAD) is a critical task in graph machine learning, with the primary objective of identifying anomalous nodes that deviate significantly from the majority. This task is widely applied in various real-world scenarios, including fraud detection and social network analysis. However, existing GAD methods still face two major challenges: (1) They are often limited to detecting anomalies in single-type interaction graphs and struggle with multiple interaction types in multiplex heterogeneous graphs. (2) In unsupervised scenarios, selecting appropriate anomaly score thresholds remains a significant challenge for accurate anomaly detection. To address the above challenges, we propose a novel Unsupervised Multiplex Graph Anomaly Detection method, named UMGAD. We first learn multi-relational correlations among nodes in multiplex heterogeneous graphs and capture anomaly information during node attribute and structure reconstruction through graph-masked autoencoder (GMAE). Then, to further extract abnormal information, we generate attribute-level and subgraph-level augmented-view graphs respectively, and perform attribute and structure reconstruction through GMAE. Finally, we learn to optimize node attributes and structural features through contrastive learning between original-view and augmented-view graphs to improve the model's ability to capture anomalies. Meanwhile, we also propose a new anomaly score threshold selection strategy, which allows the model to be independent of ground truth information in real unsupervised scenarios. Extensive experiments on four datasets show that our UMGAD significantly outperforms state-of-the-art methods, achieving average improvements of 13.48% in AUC and 11.68% in Macro-F1 across all datasets.
♻ ☆ Optimization, Isoperimetric Inequalities, and Sampling via Lyapunov Potentials
In this paper, we prove that optimizability of any F using Gradient Flow from all initializations implies a Poincar\'e Inequality for Gibbs measures mu_{beta} = e^{-\beta F}/Z at low temperature. In particular, under mild regularity assumptions on the convergence rate of Gradient Flow, we establish that mu_{beta} satisfies a Poincar\'e Inequality with constant O(C'+1/beta) for beta >= Omega(d), where C' is the Poincar\'e constant of mu_{beta} restricted to a neighborhood of the global minimizers of F. Under an additional mild condition on F, we show that mu_{beta} satisfies a Log-Sobolev Inequality with constant O(S beta C') where S denotes the second moment of mu_{beta}. Here asymptotic notation hides F-dependent parameters. At a high level, this establishes that optimizability via Gradient Flow from every initialization implies a Poincar\'e and Log-Sobolev Inequality for the low-temperature Gibbs measure, which in turn imply sampling from all initializations. Analogously, we establish that under the same assumptions, if F can be initialized from everywhere except some set S, then mu_{beta} satisfies a Weak Poincar\'e Inequality with parameters (C', mu_{beta}(S)) for \beta = Omega(d). At a high level, this shows while optimizability from 'most' initializations implies a Weak Poincar\'e Inequality, which in turn implies sampling from suitable warm starts. Our regularity assumptions are mild and as a consequence, we show we can efficiently sample from several new natural and interesting classes of non-log-concave densities, an important setting with relatively few examples. As another corollary, we obtain efficient discrete-time sampling results for log-concave measures satisfying milder regularity conditions than smoothness, similar to Lehec (2023).
comment: 52 pages. New version with new results on Weak Poincar\'e Inequalities and improved presentation
♻ ☆ Anytime-Constrained Equilibria in Polynomial Time
We extend anytime constraints to the Markov game setting and the corresponding solution concept of an anytime-constrained equilibrium (ACE). Then, we present a comprehensive theory of anytime-constrained equilibria that includes (1) a computational characterization of feasible policies, (2) a fixed-parameter tractable algorithm for computing ACE, and (3) a polynomial-time algorithm for approximately computing ACE. Since computing a feasible policy is NP-hard even for two-player zero-sum games, our approximation guarantees are optimal so long as $P \neq NP$. We also develop the first theory of efficient computation for action-constrained Markov games, which may be of independent interest.
♻ ☆ Data Distributional Properties As Inductive Bias for Systematic Generalization
Deep neural networks (DNNs) struggle at systematic generalization (SG). Several studies have evaluated the possibility to promote SG through the proposal of novel architectures, loss functions or training methodologies. Few studies, however, have focused on the role of training data properties in promoting SG. In this work, we investigate the impact of certain data distributional properties, as inductive biases for the SG ability of a multi-modal language model. To this end, we study three different properties. First, data diversity, instantiated as an increase in the possible values a latent property in the training distribution may take. Second, burstiness, where we probabilistically restrict the number of possible values of latent factors on particular inputs during training. Third, latent intervention, where a particular latent factor is altered randomly during training. We find that all three factors significantly enhance SG, with diversity contributing an 89% absolute increase in accuracy in the most affected property. Through a series of experiments, we test various hypotheses to understand why these properties promote SG. Finally, we find that Normalized Mutual Information (NMI) between latent attributes in the training distribution is strongly predictive of out-of-distribution generalization. We find that a mechanism by which lower NMI induces SG is in the geometry of representations. In particular, we find that NMI induces more parallelism in neural representations (i.e., input features coded in parallel neural vectors) of the model, a property related to the capacity of reasoning by analogy.
♻ ☆ Layer Swapping for Zero-Shot Cross-Lingual Transfer in Large Language Models ICLR 2025
Model merging, such as model souping, is the practice of combining different models with the same architecture together without further training. In this work, we present a model merging methodology that addresses the difficulty of fine-tuning Large Language Models (LLMs) for target tasks in non-English languages, where task-specific data is often unavailable. We focus on mathematical reasoning and without in-language math data, facilitate cross-lingual transfer by composing language and math capabilities. Starting from the same pretrained model, we fine-tune separate "experts" on math instruction data in English and on generic instruction data in the target language. We then replace the top and bottom transformer layers of the math expert directly with layers from the language expert, which consequently enhances math performance in the target language. The resulting merged models outperform the individual experts and other merging methods on the math benchmark, MGSM, by 10% across four major languages where math instruction data is scarce. In addition, this layer swapping is simple, inexpensive, and intuitive, as it is based on an interpretative analysis of the most important parameter changes during the fine-tuning of each expert. The ability to successfully re-compose LLMs for cross-lingual transfer in this manner opens up future possibilities to combine model expertise, create modular solutions, and transfer reasoning capabilities across languages all post hoc.
comment: ICLR 2025, Spotlight Paper, In The Thirteenth International Conference on Learning Representations, 2025
♻ ☆ Last-Iterate Convergence Properties of Regret-Matching Algorithms in Games ICLR 2025
We study last-iterate convergence properties of algorithms for solving two-player zero-sum games based on Regret Matching$^+$ (RM$^+$). Despite their widespread use for solving real games, virtually nothing is known about their last-iterate convergence. A major obstacle to analyzing RM-type dynamics is that their regret operators lack Lipschitzness and (pseudo)monotonicity. We start by showing numerically that several variants used in practice, such as RM$^+$, predictive RM$^+$ and alternating RM$^+$, all lack last-iterate convergence guarantees even on a simple $3\times 3$ matrix game. We then prove that recent variants of these algorithms based on a smoothing technique, extragradient RM$^{+}$ and smooth Predictive RM$^+$, enjoy asymptotic last-iterate convergence (without a rate), $1/\sqrt{t}$ best-iterate convergence, and when combined with restarting, linear-rate last-iterate convergence. Our analysis builds on a new characterization of the geometric structure of the limit points of our algorithms, marking a significant departure from most of the literature on last-iterate convergence. We believe that our analysis may be of independent interest and offers a fresh perspective for studying last-iterate convergence in algorithms based on non-monotone operators.
comment: Accepted to ICLR 2025
♻ ☆ Verbalized Probabilistic Graphical Modeling
Human cognition excels at transcending sensory input and forming latent representations that structure our understanding of the world. Although Large Language Models (LLMs) can produce chain-of-thought reasoning, they lack a principled framework to capture latent structures and model uncertainty, especially in compositional reasoning tasks. We propose Verbalized Probabilistic Graphical Modeling (vPGM), a Bayesian prompting framework that guides LLMs to simulate key principles of Probabilistic Graphical Models (PGMs) in natural language. Unlike many traditional probabilistic methods requiring substantial domain expertise or specialized training, vPGM bypasses expert-driven model design, making it well-suited for scenarios with limited assumptions or scarce data. We evaluated our model on several compositional reasoning tasks, both close-ended and open-ended. Our results indicate that the model effectively enhances confidence calibration and text generation quality.
♻ ☆ Decentralized Adversarial Training over Graphs
The vulnerability of machine learning models to adversarial attacks has been attracting considerable attention in recent years. Most existing studies focus on the behavior of stand-alone single-agent learners. In comparison, this work studies adversarial training over graphs, where individual agents are subjected to perturbations of varied strength levels across space. It is expected that interactions by linked agents, and the heterogeneity of the attack models that are possible over the graph, can help enhance robustness in view of the coordination power of the group. Using a min-max formulation of distributed learning, we develop a decentralized adversarial training framework for multi-agent systems. Specifically, we devise two decentralized adversarial training algorithms by relying on two popular decentralized learning strategies--diffusion and consensus. We analyze the convergence properties of the proposed framework for strongly-convex, convex, and non-convex environments, and illustrate the enhanced robustness to adversarial attacks.
comment: arXiv admin note: text overlap with arXiv:2303.01936
♻ ☆ Exploring the Potential of QEEGNet for Cross-Task and Cross-Dataset Electroencephalography Encoding with Quantum Machine Learning
Electroencephalography (EEG) is widely used in neuroscience and clinical research for analyzing brain activity. While deep learning models such as EEGNet have shown success in decoding EEG signals, they often struggle with data complexity, inter-subject variability, and noise robustness. Recent advancements in quantum machine learning (QML) offer new opportunities to enhance EEG analysis by leveraging quantum computing's unique properties. In this study, we extend the previously proposed Quantum-EEGNet (QEEGNet), a hybrid neural network incorporating quantum layers into EEGNet, to investigate its generalization ability across multiple EEG datasets. Our evaluation spans a diverse set of cognitive and motor task datasets, assessing QEEGNet's performance in different learning scenarios. Experimental results reveal that while QEEGNet demonstrates competitive performance and maintains robustness in certain datasets, its improvements over traditional deep learning methods remain inconsistent. These findings suggest that hybrid quantum-classical architectures require further optimization to fully leverage quantum advantages in EEG processing. Despite these limitations, our study provides new insights into the applicability of QML in EEG research and highlights challenges that must be addressed for future advancements.
♻ ☆ λ: A Benchmark for Data-Efficiency in Long-Horizon Indoor Mobile Manipulation Robotics
Learning to execute long-horizon mobile manipulation tasks is crucial for advancing robotics in household and workplace settings. However, current approaches are typically data-inefficient, underscoring the need for improved models that require realistically sized benchmarks to evaluate their efficiency. To address this, we introduce the LAMBDA ({\lambda}) benchmark-Long-horizon Actions for Mobile-manipulation Benchmarking of Directed Activities-which evaluates the data efficiency of models on language-conditioned, long-horizon, multi-room, multi-floor, pick-and-place tasks using a dataset of manageable size, more feasible for collection. Our benchmark includes 571 human-collected demonstrations that provide realism and diversity in simulated and real-world settings. Unlike planner-generated data, these trajectories offer natural variability and replay-verifiability, ensuring robust learning and evaluation. We leverage LAMBDA to benchmark current end-to-end learning methods and a modular neuro-symbolic approaches that combines foundation models with task and motion planning. We find that end-to-end methods-even when pretrained-yield lower success rates, while neuro-symbolic methods perform significantly better and require less data.
comment: 8 pages
♻ ☆ Linear Multidimensional Regression with Interactive Fixed-Effects
This paper studies a linear model for multidimensional panel data of three or more dimensions with unobserved interactive fixed-effects. The main estimator uses double debias methods, and requires two preliminary steps. First, the model is embedded within a two-dimensional panel framework where factor model methods in Bai (2009) lead to consistent, but slowly converging, estimates. The second step develops a weighted-within transformation that is robust to multidimensional interactive fixed-effects and achieves the parametric rate of consistency. This is combined with a double debias procedure for asymptotically normal estimates. The methods are implemented to estimate the demand elasticity for beer.
♻ ☆ AI-based association analysis for medical imaging using latent-space geometric confounder correction
This study addresses the challenges of confounding effects and interpretability in artificial-intelligence-based medical image analysis. Whereas existing literature often resolves confounding by removing confounder-related information from latent representations, this strategy risks affecting image reconstruction quality in generative models, thus limiting their applicability in feature visualization. To tackle this, we propose a different strategy that retains confounder-related information in latent representations while finding an alternative confounder-free representation of the image data. Our approach views the latent space of an autoencoder as a vector space, where imaging-related variables, such as the learning target (t) and confounder (c), have a vector capturing their variability. The confounding problem is addressed by searching a confounder-free vector which is orthogonal to the confounder-related vector but maximally collinear to the target-related vector. To achieve this, we introduce a novel correlation-based loss that not only performs vector searching in the latent space, but also encourages the encoder to generate latent representations linearly correlated with the variables. Subsequently, we interpret the confounder-free representation by sampling and reconstructing images along the confounder-free vector. The efficacy and flexibility of our proposed method are demonstrated across three applications, accommodating multiple confounders and utilizing diverse image modalities. Results affirm the method's effectiveness in reducing confounder influences, preventing wrong or misleading associations, and offering a unique visual interpretation for in-depth investigations by clinical and epidemiological researchers. The code is released in the following GitLab repository: https://gitlab.com/radiology/compopbio/ai_based_association_analysis}
comment: Accepted by Medical Image Analysis
♻ ☆ EC-DIT: Scaling Diffusion Transformers with Adaptive Expert-Choice Routing
Diffusion transformers have been widely adopted for text-to-image synthesis. While scaling these models up to billions of parameters shows promise, the effectiveness of scaling beyond current sizes remains underexplored and challenging. By explicitly exploiting the computational heterogeneity of image generations, we develop a new family of Mixture-of-Experts (MoE) models (EC-DIT) for diffusion transformers with expert-choice routing. EC-DIT learns to adaptively optimize the compute allocated to understand the input texts and generate the respective image patches, enabling heterogeneous computation aligned with varying text-image complexities. This heterogeneity provides an efficient way of scaling EC-DIT up to 97 billion parameters and achieving significant improvements in training convergence, text-to-image alignment, and overall generation quality over dense models and conventional MoE models. Through extensive ablations, we show that EC-DIT demonstrates superior scalability and adaptive compute allocation by recognizing varying textual importance through end-to-end training. Notably, in text-to-image alignment evaluation, our largest models achieve a state-of-the-art GenEval score of 71.68% and still maintain competitive inference speed with intuitive interpretability.
♻ ☆ Exploring Advanced Techniques for Visual Question Answering: A Comprehensive Comparison
Visual Question Answering (VQA) has emerged as a pivotal task in the intersection of computer vision and natural language processing, requiring models to understand and reason about visual content in response to natural language questions. Analyzing VQA datasets is essential for developing robust models that can handle the complexities of multimodal reasoning. Several approaches have been developed to examine these datasets, each offering distinct perspectives on question diversity, answer distribution, and visual-textual correlations. Despite significant progress, existing VQA models face challenges related to dataset bias, limited model complexity, commonsense reasoning gaps, rigid evaluation methods, and generalization to real world scenarios. This paper offers a detailed study of the original VQA dataset, baseline models and methods along with a comparative study of five advanced VQA models, ABC-CNN, KICNLE, Masked Vision and Language Modeling, BLIP-2, and OFA, each employing distinct methods to address these ongoing challenges.
comment: 8 pages, No figures
♻ ☆ Large Language Models are Powerful EHR Encoders
Electronic Health Records (EHRs) offer rich potential for clinical prediction, yet their inherent complexity and heterogeneity pose significant challenges for traditional machine learning approaches. Domain-specific EHR foundation models trained on large collections of unlabeled EHR data have demonstrated promising improvements in predictive accuracy and generalization; however, their training is constrained by limited access to diverse, high-quality datasets and inconsistencies in coding standards and healthcare practices. In this study, we explore the possibility of using general-purpose Large Language Models (LLMs) based embedding methods as EHR encoders. By serializing patient records into structured Markdown text, transforming codes into human-readable descriptors, we leverage the extensive generalization capabilities of LLMs pretrained on vast public corpora, thereby bypassing the need for proprietary medical datasets. We systematically evaluate two state-of-the-art LLM-embedding models, GTE-Qwen2-7B-Instruct and LLM2Vec-Llama3.1-8B-Instruct, across 15 diverse clinical prediction tasks from the EHRSHOT benchmark, comparing their performance to an EHRspecific foundation model, CLIMBR-T-Base, and traditional machine learning baselines. Our results demonstrate that LLM-based embeddings frequently match or exceed the performance of specialized models, even in few-shot settings, and that their effectiveness scales with the size of the underlying LLM and the available context window. Overall, our findings demonstrate that repurposing LLMs for EHR encoding offers a scalable and effective approach for clinical prediction, capable of overcoming the limitations of traditional EHR modeling and facilitating more interoperable and generalizable healthcare applications.
♻ ☆ VILA-U: a Unified Foundation Model Integrating Visual Understanding and Generation
VILA-U is a Unified foundation model that integrates Video, Image, Language understanding and generation. Traditional visual language models (VLMs) use separate modules for understanding and generating visual content, which can lead to misalignment and increased complexity. In contrast, VILA-U employs a single autoregressive next-token prediction framework for both tasks, eliminating the need for additional components like diffusion models. This approach not only simplifies the model but also achieves near state-of-the-art performance in visual language understanding and generation. The success of VILA-U is attributed to two main factors: the unified vision tower that aligns discrete visual tokens with textual inputs during pretraining, which enhances visual perception, and autoregressive image generation can achieve similar quality as diffusion models with high-quality dataset. This allows VILA-U to perform comparably to more complex models using a fully token-based autoregressive framework.
comment: Code: https://github.com/mit-han-lab/vila-u. The first two authors contributed equally to this work
♻ ☆ Performance of Zero-Shot Time Series Foundation Models on Cloud Data
Time series foundation models (FMs) have emerged as a popular paradigm for zero-shot multi-domain forecasting. FMs are trained on numerous diverse datasets and claim to be effective forecasters across multiple different time series domains, including cloud data. In this work we investigate this claim, exploring the effectiveness of FMs on cloud data. We demonstrate that many well-known FMs fail to generate meaningful or accurate zero-shot forecasts in this setting. We support this claim empirically, showing that FMs are outperformed consistently by simple linear baselines. We also illustrate a number of interesting pathologies, including instances where FMs suddenly output seemingly erratic, random-looking forecasts. Our results suggest a widespread failure of FMs to model cloud data.
comment: 5 pages, Preprint
♻ ☆ CVKAN: Complex-Valued Kolmogorov-Arnold Networks
In this work we propose CVKAN, a complex-valued KAN, to join the intrinsic interpretability of KANs and the advantages of Complex-Valued Neural Networks (CVNNs). We show how to transfer a KAN and the necessary associated mechanisms into the complex domain. To confirm that CVKAN meets expectations we conduct experiments on symbolic complex-valued function fitting and physically meaningful formulae as well as on a more realistic dataset from knot theory. Our proposed CVKAN is more stable and performs on par or better than real-valued KANs while requiring less parameters and a shallower network architecture, making it more explainable.
♻ ☆ Generating Graphs via Spectral Diffusion
In this paper, we present GGSD, a novel graph generative model based on 1) the spectral decomposition of the graph Laplacian matrix and 2) a diffusion process. Specifically, we propose to use a denoising model to sample eigenvectors and eigenvalues from which we can reconstruct the graph Laplacian and adjacency matrix. Using the Laplacian spectrum allows us to naturally capture the structural characteristics of the graph and work directly in the node space while avoiding the quadratic complexity bottleneck that limits the applicability of other diffusion-based methods. This, in turn, is accomplished by truncating the spectrum, which, as we show in our experiments, results in a faster yet accurate generative process, and by designing a novel transformer-based architecture linear in the number of nodes. Our permutation invariant model can also handle node features by concatenating them to the eigenvectors of each node. An extensive set of experiments on both synthetic and real-world graphs demonstrates the strengths of our model against state-of-the-art alternatives.
♻ ☆ Learning Mixtures of Gaussians Using Diffusion Models
We give a new algorithm for learning mixtures of $k$ Gaussians (with identity covariance in $\mathbb{R}^n$) to TV error $\varepsilon$, with quasi-polynomial ($O(n^{\text{poly\,log}\left(\frac{n+k}{\varepsilon}\right)})$) time and sample complexity, under a minimum weight assumption. Our results extend to continuous mixtures of Gaussians where the mixing distribution is supported on a union of $k$ balls of constant radius. In particular, this applies to the case of Gaussian convolutions of distributions on low-dimensional manifolds, or more generally sets with small covering number, for which no sub-exponential algorithm was previously known. Unlike previous approaches, most of which are algebraic in nature, our approach is analytic and relies on the framework of diffusion models. Diffusion models are a modern paradigm for generative modeling, which typically rely on learning the score function (gradient log-pdf) along a process transforming a pure noise distribution, in our case a Gaussian, to the data distribution. Despite their dazzling performance in tasks such as image generation, there are few end-to-end theoretical guarantees that they can efficiently learn nontrivial families of distributions; we give some of the first such guarantees. We proceed by deriving higher-order Gaussian noise sensitivity bounds for the score functions for a Gaussian mixture to show that that they can be inductively learned using piecewise polynomial regression (up to poly-logarithmic degree), and combine this with known convergence results for diffusion models.
♻ ☆ Regression and Forecasting of U.S. Stock Returns Based on LSTM
This paper analyses the investment returns of three stock sectors, Manuf, Hitec, and Other, in the U.S. stock market, based on the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model, in order to test the validity of the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model for the three sectors of the market. French five-factor model for the three sectors of the market. Also, the LSTM model is used to explore the additional factors affecting stock returns. The empirical results show that the Fama-French five-factor model has better validity for the three segments of the market under study, and the LSTM model has the ability to capture the factors affecting the returns of certain industries, and can better regress and predict the stock returns of the relevant industries. Keywords- Fama-French model; Carhart model; Factor model; LSTM model.
comment: 5pages
♻ ☆ A Comparative Evaluation of Quantification Methods
Quantification represents the problem of estimating the distribution of class labels on unseen data. It also represents a growing research field in supervised machine learning, for which a large variety of different algorithms has been proposed in recent years. However, a comprehensive empirical comparison of quantification methods that supports algorithm selection is not available yet. In this work, we close this research gap by conducting a thorough empirical performance comparison of 24 different quantification methods on overall more than 40 data sets, considering binary as well as multiclass quantification settings. We observe that no single algorithm generally outperforms all competitors, but identify a group of methods including the threshold selection-based Median Sweep and TSMax methods, the DyS framework including the HDy method, Forman's mixture model, and Friedman's method that performs best in the binary setting. For the multiclass setting, we observe that a different, broad group of algorithms yields good performance, including the HDx method, the Generalized Probabilistic Adjusted Count, the readme method, the energy distance minimization method, the EM algorithm for quantification, and Friedman's method. We also find that tuning the underlying classifiers has in most cases only a limited impact on the quantification performance. More generally, we find that the performance on multiclass quantification is inferior to the results obtained in the binary setting. Our results can guide practitioners who intend to apply quantification algorithms and help researchers to identify opportunities for future research.
comment: 41 pages, 18 figures, 9 tables
♻ ☆ The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
In reinforcement learning, specifying reward functions that capture the intended task can be very challenging. Reward learning aims to address this issue by learning the reward function. However, a learned reward model may have a low error on the data distribution, and yet subsequently produce a policy with large regret. We say that such a reward model has an error-regret mismatch. The main source of an error-regret mismatch is the distributional shift that commonly occurs during policy optimization. In this paper, we mathematically show that a sufficiently low expected test error of the reward model guarantees low worst-case regret, but that for any fixed expected test error, there exist realistic data distributions that allow for error-regret mismatch to occur. We then show that similar problems persist even when using policy regularization techniques, commonly employed in methods such as RLHF. We hope our results stimulate the theoretical and empirical study of improved methods to learn reward models, and better ways to measure their quality reliably.
comment: 70 pages, 4 figures
♻ ☆ Modeling Relational Patterns for Logical Query Answering over Knowledge Graphs ECAI 2024
Answering first-order logical (FOL) queries over knowledge graphs (KG) remains a challenging task mainly due to KG incompleteness. Query embedding approaches this problem by computing the low-dimensional vector representations of entities, relations, and logical queries. KGs exhibit relational patterns such as symmetry and composition and modeling the patterns can further enhance the performance of query embedding models. However, the role of such patterns in answering FOL queries by query embedding models has not been yet studied in the literature. In this paper, we fill in this research gap and empower FOL queries reasoning with pattern inference by introducing an inductive bias that allows for learning relation patterns. To this end, we develop a novel query embedding method, RoConE, that defines query regions as geometric cones and algebraic query operators by rotations in complex space. RoConE combines the advantages of Cone as a well-specified geometric representation for query embedding, and also the rotation operator as a powerful algebraic operation for pattern inference. Our experimental results on several benchmark datasets confirm the advantage of relational patterns for enhancing logical query answering task.
comment: The results reported in this paper are included in our accepted paper arXiv:2407.09212 at ECAI 2024
♻ ☆ Loss Landscape of Shallow ReLU-like Neural Networks: Stationary Points, Saddle Escape, and Network Embedding
In this paper, we study the loss landscape of one-hidden-layer neural networks with ReLU-like activation functions trained with the empirical squared loss using gradient descent (GD). We identify the stationary points of such networks, which significantly slow down loss decrease during training. To capture such points while accounting for the non-differentiability of the loss, the stationary points that we study are directional stationary points, rather than other notions like Clarke stationary points. We show that, if a stationary point does not contain "escape neurons", which are defined with first-order conditions, it must be a local minimum. Moreover, for the scalar-output case, the presence of an escape neuron guarantees that the stationary point is not a local minimum. Our results refine the description of the saddle-to-saddle training process starting from infinitesimally small (vanishing) initialization for shallow ReLU-like networks: By precluding the saddle escape types that previous works did not rule out, we advance one step closer to a complete picture of the entire dynamics. Moreover, we are also able to fully discuss how network embedding, which is to instantiate a narrower network with a wider network, reshapes the stationary points.
♻ ☆ Learning Structured Compressed Sensing with Automatic Resource Allocation
Multidimensional data acquisition often requires extensive time and poses significant challenges for hardware and software regarding data storage and processing. Rather than designing a single compression matrix as in conventional compressed sensing, structured compressed sensing yields dimension-specific compression matrices, reducing the number of optimizable parameters. Recent advances in machine learning (ML) have enabled task-based supervised learning of subsampling matrices, albeit at the expense of complex downstream models. Additionally, the sampling resource allocation across dimensions is often determined in advance through heuristics. To address these challenges, we introduce Structured COmpressed Sensing with Automatic Resource Allocation (SCOSARA) with an information theory-based unsupervised learning strategy. SCOSARA adaptively distributes samples across sampling dimensions while maximizing Fisher information content. Using ultrasound localization as a case study, we compare SCOSARA to state-of-the-art ML-based and greedy search algorithms. Simulation results demonstrate that SCOSARA can produce high-quality subsampling matrices that achieve lower Cram\'er-Rao Bound values than the baselines. In addition, SCOSARA outperforms other ML-based algorithms in terms of the number of trainable parameters, computational complexity, and memory requirements while automatically choosing the number of samples per axis.
comment: Unsupervised Learning, Information Theory, Compressed Sensing, Subsampling
♻ ☆ LADDER: Self-Improving LLMs Through Recursive Problem Decomposition
We introduce LADDER (Learning through Autonomous Difficulty-Driven Example Recursion), a framework which enables Large Language Models to autonomously improve their problem-solving capabilities through self-guided learning by recursively generating and solving progressively simpler variants of complex problems. Unlike prior approaches that require curated datasets or human feedback, LADDER leverages a model's own capabilities to generate easier question variants. We demonstrate LADDER's effectiveness in the subject of mathematical integration, improving Llama 3.2 3B's accuracy from 1% to 82% on undergraduate-level problems and enabling Qwen2.5 7B Deepseek-R1 Distilled to achieve 73% on the MIT Integration Bee qualifying examination. We also introduce TTRL (Test-Time Reinforcement Learning), where we perform reinforcement learning on variants of test problems at inference time. TTRL enables Qwen2.5 7B Deepseek-R1 Distilled to achieve a state-of-the-art score of 90% on the MIT Integration Bee qualifying examination, surpassing OpenAI o1's performance. These results show how self-directed strategic learning can achieve significant capability improvements without relying on architectural scaling or human supervision.
♻ ☆ Discovering Antagonists in Networks of Systems: Robot Deployment
A contextual anomaly detection method is proposed and applied to the physical motions of a robot swarm executing a coverage task. Using simulations of a swarm's normal behavior, a normalizing flow is trained to predict the likelihood of a robot motion within the current context of its environment. During application, the predicted likelihood of the observed motions is used by a detection criterion that categorizes a robot agent as normal or antagonistic. The proposed method is evaluated on five different strategies of antagonistic behavior. Importantly, only readily available simulated data of normal robot behavior is used for training such that the nature of the anomalies need not be known beforehand. The best detection criterion correctly categorizes at least 80% of each antagonistic type while maintaining a false positive rate of less than 5% for normal robot agents. Additionally, the method is validated in hardware experiments, yielding results similar to the simulated scenarios. Compared to the state-of-the-art approach, both the predictive performance of the normalizing flow and the robustness of the detection criterion are increased.
comment: reduced file size
♻ ☆ Hierarchical graph sampling based minibatch learning with chain preservation and variance reduction
Graph sampling based Graph Convolutional Networks (GCNs) decouple the sampling from the forward and backward propagation during minibatch training, which exhibit good scalability in terms of layer depth and graph size. We propose HIS_GCNs, a hierarchical importance graph sampling based learning method. By constructing minibatches using sampled subgraphs, HIS_GCNs gives attention to the importance of both core and periphery nodes/edges in a scale-free training graph. Specifically, it preserves the centrum of the core to most minibatches, which maintains connectivity between periphery nodes, and samples periphery edges without core node interference, in order to keep more long chains composed entirely of low-degree nodes in the same minibatch. HIS_GCNs can maximize the discrete Ricci curvature (i.e., Ollivier-Ricci curvatures) of the edges in a subgraph that enables the preservation of important chains for information propagation, and can achieve a low node embedding variance and a high convergence speed. Diverse experiments on Graph Neural Networks (GNNs) with node classification tasks confirm superior performance of HIS_GCNs in both accuracy and training time.
comment: 26 pages, 9 figures
♻ ☆ Multi-Head Self-Attending Neural Tucker Factorization
Quality-of-service (QoS) data exhibit dynamic temporal patterns that are crucial for accurately predicting missing values. These patterns arise from the evolving interactions between users and services, making it essential to capture the temporal dynamics inherent in such data for improved prediction performance. As the size and complexity of QoS datasets increase, existing models struggle to provide accurate predictions, highlighting the need for more flexible and dynamic methods to better capture the underlying patterns in large-scale QoS data. To address this issue, we introduce a neural network-based tensor factorization approach tailored for learning spatiotemporal representations of high-dimensional and incomplete (HDI) tensors, namely the Multi-head Self-attending Neural Tucker Factorization (MSNTucF). The model is elaborately designed for modeling intricate nonlinear spatiotemporal feature interaction patterns hidden in real world data with a two-fold idea. It first employs a neural network structure to generalize the traditional framework of Tucker factorization and then proposes to leverage a multi-head self-attending module to enforce nonlinear latent interaction learning. In empirical studies on two dynamic QoS datasets from real applications, the proposed MSNTucF model demonstrates superior performance compared to state-of-the-art benchmark models in estimating missing observations. This highlights its ability to learn non-linear spatiotemporal representations of HDI tensors.
♻ ☆ EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty
Autoregressive decoding makes the inference of Large Language Models (LLMs) time-consuming. In this paper, we reconsider speculative sampling and derive two key observations. Firstly, autoregression at the feature (second-to-top-layer) level is more straightforward than at the token level. Secondly, the inherent uncertainty in feature (second-to-top-layer) level autoregression constrains its performance. Based on these insights, we introduce EAGLE (Extrapolation Algorithm for Greater Language-model Efficiency), a simple yet highly efficient speculative sampling framework. By incorporating a token sequence advanced by one time step, EAGLE effectively resolves the uncertainty, enabling precise second-to-top-layer feature prediction with minimal overhead. We conducted comprehensive evaluations of EAGLE, including all models from the Vicuna and LLaMA2-Chat series, the MoE model Mixtral 8x7B Instruct, and tasks in dialogue, code generation, mathematical reasoning, and instruction following. For LLaMA2-Chat 70B, EAGLE achieved a latency speedup ratio of 2.7x-3.5x, doubled throughput, while maintaining the distribution of the generated text.
♻ ☆ Training a multilayer dynamical spintronic network with standard machine learning tools to perform time series classification
The ability to process time-series at low energy cost is critical for many applications. Recurrent neural network, which can perform such tasks, are computationally expensive when implementing in software on conventional computers. Here we propose to implement a recurrent neural network in hardware using spintronic oscillators as dynamical neurons. Using numerical simulations, we build a multi-layer network and demonstrate that we can use backpropagation through time (BPTT) and standard machine learning tools to train this network. Leveraging the transient dynamics of the spintronic oscillators, we solve the sequential digits classification task with $89.83\pm2.91~\%$ accuracy, as good as the equivalent software network. We devise guidelines on how to choose the time constant of the oscillators as well as hyper-parameters of the network to adapt to different input time scales.
comment: 7 pages, 4 figures
♻ ☆ MIXRTs: Toward Interpretable Multi-Agent Reinforcement Learning via Mixing Recurrent Soft Decision Trees
While achieving tremendous success in various fields, existing multi-agent reinforcement learning (MARL) with a black-box neural network makes decisions in an opaque manner that hinders humans from understanding the learned knowledge and how input observations influence decisions. In contrast, existing interpretable approaches usually suffer from weak expressivity and low performance. To bridge this gap, we propose MIXing Recurrent soft decision Trees (MIXRTs), a novel interpretable architecture that can represent explicit decision processes via the root-to-leaf path and reflect each agent's contribution to the team. Specifically, we construct a novel soft decision tree using a recurrent structure and demonstrate which features influence the decision-making process. Then, based on the value decomposition framework, we linearly assign credit to each agent by explicitly mixing individual action values to estimate the joint action value using only local observations, providing new insights into interpreting the cooperation mechanism. Theoretical analysis confirms that MIXRTs guarantee additivity and monotonicity in the factorization of joint action values. Evaluations on complex tasks like Spread and StarCraft II demonstrate that MIXRTs compete with existing methods while providing clear explanations, paving the way for interpretable and high-performing MARL systems.
comment: Accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence
♻ ☆ Unifying Model Predictive Path Integral Control, Reinforcement Learning, and Diffusion Models for Optimal Control and Planning
Model Predictive Path Integral (MPPI) control, Reinforcement Learning (RL), and Diffusion Models have each demonstrated strong performance in trajectory optimization, decision-making, and motion planning. However, these approaches have traditionally been treated as distinct methodologies with separate optimization frameworks. In this work, we establish a unified perspective that connects MPPI, RL, and Diffusion Models through gradient-based optimization on the Gibbs measure. We first show that MPPI can be interpreted as performing gradient ascent on a smoothed energy function. We then demonstrate that Policy Gradient methods reduce to MPPI by applying an exponential transformation to the objective function. Additionally, we establish that the reverse sampling process in diffusion models follows the same update rule as MPPI.
comment: updated RL subsection in Main section
♻ ☆ A Unifying Framework for Learning Argumentation Semantics
Argumentation is a very active research field of Artificial Intelligence concerned with the representation and evaluation of arguments used in dialogues between humans and/or artificial agents. Acceptability semantics of formal argumentation systems define the criteria for the acceptance or rejection of arguments. Several software systems, known as argumentation solvers, have been developed to compute the accepted/rejected arguments using such criteria. These include systems that learn to identify the accepted arguments using non-interpretable methods. In this paper we present a novel framework, which uses an Inductive Logic Programming approach to learn the acceptability semantics for several abstract and structured argumentation frameworks in an interpretable way. Through an empirical evaluation we show that our framework outperforms existing argumentation solvers, thus opening up new future research directions in the area of formal argumentation and human-machine dialogues.
♻ ☆ Let the Code LLM Edit Itself When You Edit the Code ICLR 2025
In this work, we investigate a typical scenario in code generation where a developer edits existing code in real time and requests a code assistant, e.g., a large language model, to re-predict the next token or next line on the fly. Naively, the LLM needs to re-encode the entire KV cache to provide an accurate prediction. However, this process is computationally expensive, especially when the sequence length is long. Simply encoding the edited subsequence and integrating it to the original KV cache meets the temporal confusion problem, leading to significantly worse performance. We address this efficiency and accuracy trade-off by introducing \underline{\textbf{Positional \textbf{I}ntegrity \textbf{E}ncoding} (PIE). Building upon the rotary positional encoding, PIE first removes the rotary matrices in the Key cache that introduce temporal confusion and then reapplies the correct rotary matrices. This process ensures that positional relationships between tokens are correct and requires only a single round of matrix multiplication. We validate the effectiveness of PIE through extensive experiments on the RepoBench-C-8k dataset, utilizing DeepSeek-Coder models with 1.3B, 6.7B, and 33B parameters. Our evaluation includes three real-world coding tasks: code insertion, code deletion, and multi-place code editing. Results demonstrate that PIE reduces computational overhead by over 85% compared to the standard full recomputation approach across all model sizes and tasks while well approximating the model performance.
comment: ICLR 2025 Camera Ready
♻ ☆ Dual Consolidation for Pre-Trained Model-Based Domain-Incremental Learning CVPR 2025
Domain-Incremental Learning (DIL) involves the progressive adaptation of a model to new concepts across different domains. While recent advances in pre-trained models provide a solid foundation for DIL, learning new concepts often results in the catastrophic forgetting of pre-trained knowledge. Specifically, sequential model updates can overwrite both the representation and the classifier with knowledge from the latest domain. Thus, it is crucial to develop a representation and corresponding classifier that accommodate all seen domains throughout the learning process. To this end, we propose DUal ConsolidaTion (Duct) to unify and consolidate historical knowledge at both the representation and classifier levels. By merging the backbone of different stages, we create a representation space suitable for multiple domains incrementally. The merged representation serves as a balanced intermediary that captures task-specific features from all seen domains. Additionally, to address the mismatch between consolidated embeddings and the classifier, we introduce an extra classifier consolidation process. Leveraging class-wise semantic information, we estimate the classifier weights of old domains within the latest embedding space. By merging historical and estimated classifiers, we align them with the consolidated embedding space, facilitating incremental classification. Extensive experimental results on four benchmark datasets demonstrate Duct's state-of-the-art performance. Code is available at https://github.com/Estrella-fugaz/CVPR25-Duct
comment: Accepted to CVPR 2025. Code is available at https://github.com/Estrella-fugaz/CVPR25-Duct
♻ ☆ SYN-LUNGS: Towards Simulating Lung Nodules with Anatomy-Informed Digital Twins for AI Training
AI models for lung cancer screening are limited by data scarcity, impacting generalizability and clinical applicability. Generative models address this issue but are constrained by training data variability. We introduce SYN-LUNGS, a framework for generating high-quality 3D CT images with detailed annotations. SYN-LUNGS integrates XCAT3 phantoms for digital twin generation, X-Lesions for nodule simulation (varying size, location, and appearance), and DukeSim for CT image formation with vendor and parameter variability. The dataset includes 3,072 nodule images from 1,044 simulated CT scans, with 512 lesions and 174 digital twins. Models trained on clinical + simulated data outperform clinical only models, achieving 10% improvement in detection, 2-9% in segmentation and classification, and enhanced synthesis.By incorporating anatomy-informed simulations, SYN-LUNGS provides a scalable approach for AI model development, particularly in rare disease representation and improving model reliability.
comment: 6 figures, 12 pages
♻ ☆ Balancing Efficiency and Effectiveness: An LLM-Infused Approach for Optimized CTR Prediction
Click-Through Rate (CTR) prediction is essential in online advertising, where semantic information plays a pivotal role in shaping user decisions and enhancing CTR effectiveness. Capturing and modeling deep semantic information, such as a user's preference for "H\"aagen-Dazs' HEAVEN strawberry light ice cream" due to its health-conscious and premium attributes, is challenging. Traditional semantic modeling often overlooks these intricate details at the user and item levels. To bridge this gap, we introduce a novel approach that models deep semantic information end-to-end, leveraging the comprehensive world knowledge capabilities of Large Language Models (LLMs). Our proposed LLM-infused CTR prediction framework(Multi-level Deep Semantic Information Infused CTR model via Distillation, MSD) is designed to uncover deep semantic insights by utilizing LLMs to extract and distill critical information into a smaller, more efficient model, enabling seamless end-to-end training and inference. Importantly, our framework is carefully designed to balance efficiency and effectiveness, ensuring that the model not only achieves high performance but also operates with optimal resource utilization. Online A/B tests conducted on the Meituan sponsored-search system demonstrate that our method significantly outperforms baseline models in terms of Cost Per Mile (CPM) and CTR, validating its effectiveness, scalability, and balanced approach in real-world applications.
comment: 5 pages, 4 figures,4 tables
♻ ☆ Structured IB: Improving Information Bottleneck with Structured Feature Learning
The Information Bottleneck (IB) principle has emerged as a promising approach for enhancing the generalization, robustness, and interpretability of deep neural networks, demonstrating efficacy across image segmentation, document clustering, and semantic communication. Among IB implementations, the IB Lagrangian method, employing Lagrangian multipliers, is widely adopted. While numerous methods for the optimizations of IB Lagrangian based on variational bounds and neural estimators are feasible, their performance is highly dependent on the quality of their design, which is inherently prone to errors. To address this limitation, we introduce Structured IB, a framework for investigating potential structured features. By incorporating auxiliary encoders to extract missing informative features, we generate more informative representations. Our experiments demonstrate superior prediction accuracy and task-relevant information preservation compared to the original IB Lagrangian method, even with reduced network size.
♻ ☆ Revisiting Random Walks for Learning on Graphs
We revisit a simple model class for machine learning on graphs, where a random walk on a graph produces a machine-readable record, and this record is processed by a deep neural network to directly make vertex-level or graph-level predictions. We call these stochastic machines random walk neural networks (RWNNs), and through principled analysis, show that we can design them to be isomorphism invariant while capable of universal approximation of graph functions in probability. A useful finding is that almost any kind of record of random walks guarantees probabilistic invariance as long as the vertices are anonymized. This enables us, for example, to record random walks in plain text and adopt a language model to read these text records to solve graph tasks. We further establish a parallelism to message passing neural networks using tools from Markov chain theory, and show that over-smoothing in message passing is alleviated by construction in RWNNs, while over-squashing manifests as probabilistic under-reaching. We empirically demonstrate RWNNs on a range of problems, verifying our theoretical analysis and demonstrating the use of language models for separating strongly regular graphs where 3-WL test fails, and transductive classification on arXiv citation network. Code is available at https://github.com/jw9730/random-walk.
comment: 51 pages, 14 figures
♻ ☆ Position: Don't use the CLT in LLM evals with fewer than a few hundred datapoints
Rigorous statistical evaluations of large language models (LLMs), including valid error bars and significance testing, are essential for meaningful and reliable performance assessment. Currently, when such statistical measures are reported, they typically rely on the Central Limit Theorem (CLT). In this position paper, we argue that while CLT-based methods for uncertainty quantification are appropriate when benchmarks consist of thousands of examples, they fail to provide adequate uncertainty estimates for LLM evaluations that rely on smaller, highly specialized benchmarks. In these small-data settings, we demonstrate that CLT-based methods perform very poorly, usually dramatically underestimating uncertainty (i.e. producing error bars that are too small). We give recommendations for alternative frequentist and Bayesian methods that are both easy to implement and more appropriate in these increasingly common scenarios. We provide a simple Python library for these Bayesian methods at https://github.com/sambowyer/bayes_evals .
comment: 36 pages, 37 figures
♻ ☆ How Ambiguous Are the Rationales for Natural Language Reasoning? A Simple Approach to Handling Rationale Uncertainty COLING2025
The quality of rationales is essential in the reasoning capabilities of language models. Rationales not only enhance reasoning performance in complex natural language tasks but also justify model decisions. However, obtaining impeccable rationales is often impossible. Our study aims to investigate how ambiguous rationales play in model performances of natural language reasoning. We first assess the ambiguity of rationales through the lens of entropy and uncertainty in model prior beliefs, exploring its impact on task performance. We then propose a simple way to guide models to choose between two different reasoning paths depending on the ambiguity of the rationales. Our empirical results demonstrate that this approach leads to robust performance, particularly in adversarial scenarios where rationale quality is inconsistent.
comment: Accepted to COLING2025
♻ ☆ From Lazy to Rich: Exact Learning Dynamics in Deep Linear Networks
Biological and artificial neural networks develop internal representations that enable them to perform complex tasks. In artificial networks, the effectiveness of these models relies on their ability to build task specific representation, a process influenced by interactions among datasets, architectures, initialization strategies, and optimization algorithms. Prior studies highlight that different initializations can place networks in either a lazy regime, where representations remain static, or a rich/feature learning regime, where representations evolve dynamically. Here, we examine how initialization influences learning dynamics in deep linear neural networks, deriving exact solutions for lambda-balanced initializations-defined by the relative scale of weights across layers. These solutions capture the evolution of representations and the Neural Tangent Kernel across the spectrum from the rich to the lazy regimes. Our findings deepen the theoretical understanding of the impact of weight initialization on learning regimes, with implications for continual learning, reversal learning, and transfer learning, relevant to both neuroscience and practical applications.
comment: 10 pages, 8 figures
♻ ☆ Generative modeling through internal high-dimensional chaotic activity
Generative modeling aims at producing new datapoints whose statistical properties resemble the ones in a training dataset. In recent years, there has been a burst of machine learning techniques and settings that can achieve this goal with remarkable performances. In most of these settings, one uses the training dataset in conjunction with noise, which is added as a source of statistical variability and is essential for the generative task. Here, we explore the idea of using internal chaotic dynamics in high-dimensional chaotic systems as a way to generate new datapoints from a training dataset. We show that simple learning rules can achieve this goal within a set of vanilla architectures and characterize the quality of the generated datapoints through standard accuracy measures.
♻ ☆ D$^2$-DPM: Dual Denoising for Quantized Diffusion Probabilistic Models AAAI2025
Diffusion models have achieved cutting-edge performance in image generation. However, their lengthy denoising process and computationally intensive score estimation network impede their scalability in low-latency and resource-constrained scenarios. Post-training quantization (PTQ) compresses and accelerates diffusion models without retraining, but it inevitably introduces additional quantization noise, resulting in mean and variance deviations. In this work, we propose D2-DPM, a dual denoising mechanism aimed at precisely mitigating the adverse effects of quantization noise on the noise estimation network. Specifically, we first unravel the impact of quantization noise on the sampling equation into two components: the mean deviation and the variance deviation. The mean deviation alters the drift coefficient of the sampling equation, influencing the trajectory trend, while the variance deviation magnifies the diffusion coefficient, impacting the convergence of the sampling trajectory. The proposed D2-DPM is thus devised to denoise the quantization noise at each time step, and then denoise the noisy sample through the inverse diffusion iterations. Experimental results demonstrate that D2-DPM achieves superior generation quality, yielding a 1.42 lower FID than the full-precision model while achieving 3.99x compression and 11.67x bit-operation acceleration.
comment: 9 pages, 4 figures, acceptted by AAAI2025, the code is available at https://github.com/taylorjocelyn/d2-dpm
♻ ☆ Comet: Fine-grained Computation-communication Overlapping for Mixture-of-Experts
Mixture-of-experts (MoE) has been extensively employed to scale large language models to trillion-plus parameters while maintaining a fixed computational cost. The development of large MoE models in the distributed scenario encounters the problem of large communication overhead. The inter-device communication of a MoE layer can occupy 47% time of the entire model execution with popular models and frameworks. Therefore, existing methods suggest the communication in a MoE layer to be pipelined with the computation for overlapping. However, these coarse grained overlapping schemes introduce a notable impairment of computational efficiency and the latency concealing is sub-optimal. To this end, we present COMET, an optimized MoE system with fine-grained communication-computation overlapping. Leveraging data dependency analysis and task rescheduling, COMET achieves precise fine-grained overlapping of communication and computation. Through adaptive workload assignment, COMET effectively eliminates fine-grained communication bottlenecks and enhances its adaptability across various scenarios. Our evaluation shows that COMET accelerates the execution of a single MoE layer by $1.96\times$ and for end-to-end execution, COMET delivers a $1.71\times$ speedup on average. COMET has been adopted in the production environment of clusters with ten-thousand-scale of GPUs, achieving savings of millions of GPU hours.
♻ ☆ Scalable Multi-Agent Reinforcement Learning for Residential Load Scheduling under Data Governance
As a data-driven approach, multi-agent reinforcement learning (MARL) has made remarkable advances in solving cooperative residential load scheduling problems. However, centralized training, the most common paradigm for MARL, limits large-scale deployment in communication-constrained cloud-edge environments. As a remedy, distributed training shows unparalleled advantages in real-world applications but still faces challenge with system scalability, e.g., the high cost of communication overhead during coordinating individual agents, and needs to comply with data governance in terms of privacy. In this work, we propose a novel MARL solution to address these two practical issues. Our proposed approach is based on actor-critic methods, where the global critic is a learned function of individual critics computed solely based on local observations of households. This scheme preserves household privacy completely and significantly reduces communication cost. Simulation experiments demonstrate that the proposed framework achieves comparable performance to the state-of-the-art actor-critic framework without data governance and communication constraints.
♻ ☆ Assessing Robustness via Score-Based Adversarial Image Generation
Most adversarial attacks and defenses focus on perturbations within small $\ell_p$-norm constraints. However, $\ell_p$ threat models cannot capture all relevant semantics-preserving perturbations, and hence, the scope of robustness evaluations is limited. In this work, we introduce Score-Based Adversarial Generation (ScoreAG), a novel framework that leverages the advancements in score-based generative models to generate unrestricted adversarial examples that overcome the limitations of $\ell_p$-norm constraints. Unlike traditional methods, ScoreAG maintains the core semantics of images while generating adversarial examples, either by transforming existing images or synthesizing new ones entirely from scratch. We further exploit the generative capability of ScoreAG to purify images, empirically enhancing the robustness of classifiers. Our extensive empirical evaluation demonstrates that ScoreAG improves upon the majority of state-of-the-art attacks and defenses across multiple benchmarks. This work highlights the importance of investigating adversarial examples bounded by semantics rather than $\ell_p$-norm constraints. ScoreAG represents an important step towards more encompassing robustness assessments.
♻ ☆ Utilizing Sequential Information of General Lab-test Results and Diagnoses History for Differential Diagnosis of Dementia
Early diagnosis of Alzheimer's Disease (AD) faces multiple data-related challenges, including high variability in patient data, limited access to specialized diagnostic tests, and overreliance on single-type indicators. These challenges are exacerbated by the progressive nature of AD, where subtle pathophysiological changes often precede clinical symptoms by decades. To address these limitations, this study proposes a novel approach that takes advantage of routinely collected general laboratory test histories for the early detection and differential diagnosis of AD. By modeling lab test sequences as "sentences", we apply word embedding techniques to capture latent relationships between tests and employ deep time series models, including long-short-term memory (LSTM) and Transformer networks, to model temporal patterns in patient records. Experimental results demonstrate that our approach improves diagnostic accuracy and enables scalable and costeffective AD screening in diverse clinical settings.
comment: 7 pages, 6 figures. This work has been submitted to the Elsevier for possible publication
♻ ☆ Thermodynamic Computing via Autonomous Quantum Thermal Machines
We develop a physics-based model for classical computation based on autonomous quantum thermal machines. These machines consist of few interacting quantum bits (qubits) connected to several environments at different temperatures. Heat flows through the machine are here exploited for computing. The process starts by setting the temperatures of the environments according to the logical input. The machine evolves, eventually reaching a non-equilibrium steady state, from which the output of the computation can be determined via the temperature of an auxilliary finite-size reservoir. Such a machine, which we term a ``thermodynamic neuron'', can implement any linearly-separable function, and we discuss explicitly the cases of NOT, 3-MAJORITY and NOR gates. In turn, we show that a network of thermodynamic neurons can perform any desired function. We discuss the close connection between our model and artificial neurons (perceptrons), and argue that our model provides an alternative physics-based analogue implementation of neural networks, and more generally a platform for thermodynamic computing.
comment: 15 + 5 pages. Published version
♻ ☆ Nonparametric Control Koopman Operators
This paper presents a novel Koopman (composition) operator representation framework for control systems in reproducing kernel Hilbert spaces (RKHSs) that is free of explicit dictionary or input parametrizations. By establishing fundamental equivalences between different model representations, we are able to close the gap of control system operator learning and infinite-dimensional regression, enabling various empirical estimators and the connection to well-understood learning theory in RKHSs under one unified framework. As a consequence, our proposed framework allows for arbitrary accurate finite-rank approximations in infinite-dimensional spaces and leads to finite-dimensional predictors without apriori restrictions to a finite span of functions or inputs. To enable applications to high-dimensional control systems, we improve the scalability of our proposed control Koopman operator estimates by utilizing sketching techniques. Numerical experiments demonstrate superior prediction accuracy compared to bilinear EDMD, especially in high dimensions. Finally, we show that our learned models are readily interfaced with linear-parameter-varying techniques for model predictive control.
comment: The authors' electronic preprint version of an article submitted to IEEE for publication
♻ ☆ Geometry-aware RL for Manipulation of Varying Shapes and Deformable Objects ICLR 2025
Manipulating objects with varying geometries and deformable objects is a major challenge in robotics. Tasks such as insertion with different objects or cloth hanging require precise control and effective modelling of complex dynamics. In this work, we frame this problem through the lens of a heterogeneous graph that comprises smaller sub-graphs, such as actuators and objects, accompanied by different edge types describing their interactions. This graph representation serves as a unified structure for both rigid and deformable objects tasks, and can be extended further to tasks comprising multiple actuators. To evaluate this setup, we present a novel and challenging reinforcement learning benchmark, including rigid insertion of diverse objects, as well as rope and cloth manipulation with multiple end-effectors. These tasks present a large search space, as both the initial and target configurations are uniformly sampled in 3D space. To address this issue, we propose a novel graph-based policy model, dubbed Heterogeneous Equivariant Policy (HEPi), utilizing $SE(3)$ equivariant message passing networks as the main backbone to exploit the geometric symmetry. In addition, by modeling explicit heterogeneity, HEPi can outperform Transformer-based and non-heterogeneous equivariant policies in terms of average returns, sample efficiency, and generalization to unseen objects. Our project page is available at https://thobotics.github.io/hepi.
comment: Accepted at ICLR 2025 (Oral)
♻ ☆ Compare different SG-Schemes based on large least square problems
This study reviews popular stochastic gradient-based schemes based on large least-square problems. These schemes, often called optimizers in machine learning, play a crucial role in finding better model parameters. Hence, this study focuses on viewing such optimizers with different hyper-parameters and analyzing them based on least square problems. Codes that produced results in this work are available on https://github.com/q-viper/gradients-based-methods-on-large-least-square.
♻ ☆ A Simple and Effective Reinforcement Learning Method for Text-to-Image Diffusion Fine-tuning
Reinforcement learning ( RL)-based fine-tuning has emerged as a powerful approach for aligning diffusion models with black-box objectives. Proximal policy optimization (PPO) is the most popular choice of method for policy optimization. While effective in terms of performance, PPO is highly sensitive to hyper-parameters and involves substantial computational overhead. REINFORCE, on the other hand, mitigates some computational complexities such as high memory overhead and sensitive hyper-parameter tuning, but has suboptimal performance due to high-variance and sample inefficiency. While the variance of the REINFORCE can be reduced by sampling multiple actions per input prompt and using a baseline correction term, it still suffers from sample inefficiency. To address these challenges, we systematically analyze the efficiency-effectiveness trade-off between REINFORCE and PPO, and propose leave-one-out PPO ( LOOP), a novel RL for diffusion fine-tuning method. LOOP combines variance reduction techniques from REINFORCE, such as sampling multiple actions per input prompt and a baseline correction term, with the robustness and sample efficiency of PPO via clipping and importance sampling. Our results demonstrate that LOOP effectively improves diffusion models on various black-box objectives, and achieves a better balance between computational efficiency and performance.
♻ ☆ A new Linear Time Bi-level $\ell_{1,\infty}$ projection ; Application to the sparsification of auto-encoders neural networks
The $\ell_{1,\infty}$ norm is an efficient-structured projection, but the complexity of the best algorithm is, unfortunately, $\mathcal{O}\big(n m \log(n m)\big)$ for a matrix $n\times m$.\\ In this paper, we propose a new bi-level projection method, for which we show that the time complexity for the $\ell_{1,\infty}$ norm is only $\mathcal{O}\big(n m \big)$ for a matrix $n\times m$. Moreover, we provide a new $\ell_{1,\infty}$ identity with mathematical proof and experimental validation. Experiments show that our bi-level $\ell_{1,\infty}$ projection is $2.5$ times faster than the actual fastest algorithm and provides the best sparsity while keeping the same accuracy in classification applications.
comment: arXiv admin note: substantial text overlap with arXiv:2405.02086
♻ ☆ Neural Interpretable Reasoning
We formalize a novel modeling framework for achieving interpretability in deep learning, anchored in the principle of inference equivariance. While the direct verification of interpretability scales exponentially with the number of variables of the system, we show that this complexity can be mitigated by treating interpretability as a Markovian property and employing neural re-parametrization techniques. Building on these insights, we propose a new modeling paradigm -- neural generation and interpretable execution -- that enables scalable verification of equivariance. This paradigm provides a general approach for designing Neural Interpretable Reasoners that are not only expressive but also transparent.
♻ ☆ Revisiting Weak-to-Strong Generalization in Theory and Practice: Reverse KL vs. Forward KL
As large language models advance toward superhuman performance, ensuring their alignment with human values and abilities grows increasingly complex. Weak-to-strong generalization offers a promising approach by leveraging predictions from weaker models to guide stronger systems, but its effectiveness could be constrained by the inherent noise and inaccuracies in these weak predictions. To address this, we propose a theoretically grounded approach that replaces forward KL divergence-whose mass-covering behavior risks overfitting to imperfect weak signals-with reverse KL divergence. Reverse KL divergence's zero-forcing effect prioritizes high-confidence predictions, effectively mitigating the influence of unreliable weak supervision. Theoretically, we extend existing bounds and derive tighter lower bounds for both forward and reverse KL divergence, establishing that reverse KL achieves at least comparable guarantees to forward KL. Notably, when a sufficiently pre-trained strong model is fine-tuned on the last linear layer, reverse KL guarantees that it outperforms its weak supervisor by the magnitude of their disagreement. Empirically, we demonstrate that reverse KL and reverse cross-entropy enable strong models to successfully outperform those trained with forward KL and standard cross-entropy across most settings, highlighting the practical advantages of these reverse losses.
♻ ☆ Satellite image classification with neural quantum kernels
Achieving practical applications of quantum machine learning for real-world scenarios remains challenging despite significant theoretical progress. This paper proposes a novel approach for classifying satellite images, a task of particular relevance to the earth observation (EO) industry, using quantum machine learning techniques. Specifically, we focus on classifying images that contain solar panels, addressing a complex real-world classification problem. Our approach begins with classical pre-processing to reduce the dimensionality of the satellite image dataset. We then apply neural quantum kernels (NQKs)-quantum kernels derived from trained quantum neural networks (QNNs)-for classification. We evaluate several strategies within this framework, demonstrating results that are competitive with the best classical methods. Key findings include the robustness of or results and their scalability, with successful performance achieved up to 8 qubits.
♻ ☆ M2Lingual: Enhancing Multilingual, Multi-Turn Instruction Alignment in Large Language Models
Instruction finetuning (IFT) is critical for aligning Large Language Models (LLMs) to follow instructions. While many effective IFT datasets have been introduced recently, they predominantly focus on high-resource languages like English. To better align LLMs across a broad spectrum of languages and tasks, we propose a fully synthetic, novel taxonomy (Evol) guided Multilingual, Multi-turn instruction finetuning dataset, called M2Lingual. It is constructed by first selecting a diverse set of seed examples and then utilizing the proposed Evol taxonomy to convert these seeds into complex and challenging multi-turn instructions. We demonstrate the effectiveness of M2Lingual by training LLMs of varying sizes and showcasing the enhanced performance across a diverse set of languages. We contribute the 2 step Evol taxonomy with the guided generation code: https://github.com/ServiceNow/M2Lingual, as well as the first fully synthetic, general and task-oriented, multi-turn, multilingual dataset built with Evol - M2Lingual: https://huggingface.co/datasets/ServiceNow-AI/ M2Lingual - containing 182K total IFT pairs, covering 70 languages and 17+ NLP tasks.
comment: 39 pages
♻ ☆ Time series classification with random convolution kernels: pooling operators and input representations matter
This article presents a new approach based on MiniRocket, called SelF-Rocket, for fast time series classification (TSC). Unlike existing approaches based on random convolution kernels, it dynamically selects the best couple of input representations and pooling operator during the training process. SelF-Rocket achieves state-of-the-art accuracy on the University of California Riverside (UCR) TSC benchmark datasets.
comment: v1: initial version, incorrect evaluation. v2: Method improved, evaluation corrected, title simplified
♻ ☆ SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity Reduction ICLR 2025
Large Language Models (LLMs) have demonstrated improved generation performance by incorporating externally retrieved knowledge, a process known as retrieval-augmented generation (RAG). Despite the potential of this approach, existing studies evaluate RAG effectiveness by 1) assessing retrieval and generation components jointly, which obscures retrieval's distinct contribution, or 2) examining retrievers using traditional metrics such as NDCG, which creates a gap in understanding retrieval's true utility in the overall generation process. To address the above limitations, in this work, we introduce an automatic evaluation method that measures retrieval quality through the lens of information gain within the RAG framework. Specifically, we propose Semantic Perplexity (SePer), a metric that captures the LLM's internal belief about the correctness of the retrieved information. We quantify the utility of retrieval by the extent to which it reduces semantic perplexity post-retrieval. Extensive experiments demonstrate that SePer not only aligns closely with human preferences but also offers a more precise and efficient evaluation of retrieval utility across diverse RAG scenarios.
comment: ICLR 2025 Spotlight
♻ ☆ FABG : End-to-end Imitation Learning for Embodied Affective Human-Robot Interaction
This paper proposes FABG (Facial Affective Behavior Generation), an end-to-end imitation learning system for human-robot interaction, designed to generate natural and fluid facial affective behaviors. In interaction, effectively obtaining high-quality demonstrations remains a challenge. In this work, we develop an immersive virtual reality (VR) demonstration system that allows operators to perceive stereoscopic environments. This system ensures "the operator's visual perception matches the robot's sensory input" and "the operator's actions directly determine the robot's behaviors" - as if the operator replaces the robot in human interaction engagements. We propose a prediction-driven latency compensation strategy to reduce robotic reaction delays and enhance interaction fluency. FABG naturally acquires human interactive behaviors and subconscious motions driven by intuition, eliminating manual behavior scripting. We deploy FABG on a real-world 25-degree-of-freedom (DoF) humanoid robot, validating its effectiveness through four fundamental interaction tasks: expression response, dynamic gaze, foveated attention, and gesture recognition, supported by data collection and policy training. Project website: https://cybergenies.github.io
comment: Project website: https://cybergenies.github.io
♻ ☆ Balanced Mixed-Type Tabular Data Synthesis with Diffusion Models
Diffusion models have emerged as a robust framework for various generative tasks, including tabular data synthesis. However, current tabular diffusion models tend to inherit bias in the training dataset and generate biased synthetic data, which may influence discriminatory actions. In this research, we introduce a novel tabular diffusion model that incorporates sensitive guidance to generate fair synthetic data with balanced joint distributions of the target label and sensitive attributes, such as sex and race. The empirical results demonstrate that our method effectively mitigates bias in training data while maintaining the quality of the generated samples. Furthermore, we provide evidence that our approach outperforms existing methods for synthesizing tabular data on fairness metrics such as demographic parity ratio and equalized odds ratio, achieving improvements of over $10\%$. Our implementation is available at https://github.com/comp-well-org/fair-tab-diffusion.
comment: OpenReview: https://openreview.net/forum?id=dvRysCqmYQ
♻ ☆ VideoRAG: Retrieval-Augmented Generation over Video Corpus
Retrieval-Augmented Generation (RAG) is a powerful strategy for improving the factual accuracy of models by retrieving external knowledge relevant to queries and incorporating it into the generation process. However, existing approaches primarily focus on text, with some recent advancements considering images, and they largely overlook videos, a rich source of multimodal knowledge capable of representing contextual details more effectively than any other modality. While very recent studies explore the use of videos in response generation, they either predefine query-associated videos without retrieval or convert videos into textual descriptions losing multimodal richness. To tackle these, we introduce VideoRAG, a framework that not only dynamically retrieves videos based on their relevance with queries but also utilizes both visual and textual information. The operation of VideoRAG is powered by recent Large Video Language Models (LVLMs), which enable the direct processing of video content to represent it for retrieval and the seamless integration of retrieved videos jointly with queries for response generation. Also, inspired by that the context size of LVLMs may not be sufficient to process all frames in extremely long videos and not all frames are equally important, we introduce a video frame selection mechanism to extract the most informative subset of frames, along with a strategy to extract textual information from videos (as it can aid the understanding of video content) when their subtitles are not available. We experimentally validate the effectiveness of VideoRAG, showcasing that it is superior to relevant baselines. Code is available at https://github.com/starsuzi/VideoRAG.
♻ ☆ Analysing Multiscale Clusterings with Persistent Homology
In data clustering, it is often desirable to find not just a single partition into clusters but a sequence of partitions that describes the data at different scales (or levels of coarseness). A natural problem then is to analyse and compare the (not necessarily hierarchical) sequences of partitions that underpin such multiscale descriptions. Here, we use tools from topological data analysis and introduce the Multiscale Clustering Filtration (MCF), a well-defined and stable filtration of abstract simplicial complexes that encodes arbitrary cluster assignments in a sequence of partitions across scales of increasing coarseness. We show that the zero-dimensional persistent homology of the MCF measures the degree of hierarchy of this sequence, and the higher-dimensional persistent homology tracks the emergence and resolution of conflicts between cluster assignments across the sequence of partitions. To broaden the theoretical foundations of the MCF, we provide an equivalent construction via a nerve complex filtration, and we show that, in the hierarchical case, the MCF reduces to a Vietoris-Rips filtration of an ultrametric space. Using synthetic data, we then illustrate how the persistence diagram of the MCF provides a feature map that can serve to characterise and classify multiscale clusterings.
comment: This work was presented at the Dagstuhl Seminar (23192) on "Topological Data Analysis and Applications"
♻ ☆ A Survey on Quantum Machine Learning: Current Trends, Challenges, Opportunities, and the Road Ahead
Quantum Computing (QC) claims to improve the efficiency of solving complex problems, compared to classical computing. When QC is integrated with Machine Learning (ML), it creates a Quantum Machine Learning (QML) system. This paper aims to provide a thorough understanding of the foundational concepts of QC and its notable advantages over classical computing. Following this, we delve into the key aspects of QML in a detailed and comprehensive manner. In this survey, we investigate a variety of QML algorithms, discussing their applicability across different domains. We examine quantum datasets, highlighting their unique characteristics and advantages. The survey also covers the current state of hardware technologies, providing insights into the latest advancements and their implications for QML. Additionally, we review the software tools and simulators available for QML development, discussing their features and usability. Furthermore, we explore practical applications of QML, illustrating how it can be leveraged to solve real-world problems more efficiently than classical ML methods. This paper serves as a valuable resource for readers seeking to understand the current state-of-the-art techniques in the QML field, offering a solid foundation to embark on further exploration and development in this rapidly evolving area.
♻ ☆ Open-World Reinforcement Learning over Long Short-Term Imagination ICLR 2025
Training visual reinforcement learning agents in a high-dimensional open world presents significant challenges. While various model-based methods have improved sample efficiency by learning interactive world models, these agents tend to be "short-sighted", as they are typically trained on short snippets of imagined experiences. We argue that the primary challenge in open-world decision-making is improving the exploration efficiency across a vast state space, especially for tasks that demand consideration of long-horizon payoffs. In this paper, we present LS-Imagine, which extends the imagination horizon within a limited number of state transition steps, enabling the agent to explore behaviors that potentially lead to promising long-term feedback. The foundation of our approach is to build a $\textit{long short-term world model}$. To achieve this, we simulate goal-conditioned jumpy state transitions and compute corresponding affordance maps by zooming in on specific areas within single images. This facilitates the integration of direct long-term values into behavior learning. Our method demonstrates significant improvements over state-of-the-art techniques in MineDojo.
comment: Accepted by ICLR 2025 Oral. Project page: https://qiwang067.github.io/ls-imagine
♻ ☆ Danoliteracy of Generative Large Language Models
The language technology moonshot moment of Generative Large Language Models (GLLMs) was not limited to English: These models brought a surge of technological applications, investments, and hype to low-resource languages as well. However, the capabilities of these models in languages such as Danish were, until recently, difficult to verify beyond qualitative demonstrations due to a lack of applicable evaluation corpora. We present a GLLM benchmark to evaluate \emph{Danoliteracy}, a measure of Danish language and cultural competency across eight diverse scenarios such as Danish citizenship tests and abstractive social media question answering. This limited-size benchmark was found to produce a robust ranking that correlates to human feedback at $\rho \sim 0.8$ with GPT-4 and Claude Opus models achieving the highest rankings. Analyzing these model results across scenarios, we find one strong underlying factor explaining $95\%$ of scenario performance variance for GLLMs in Danish, suggesting a $g$ factor of model consistency in language adaptation.
comment: 16 pages, 13 figures, Accepted to NoDaLiDa/Baltic-HLT 2025
♻ ☆ SelaFD:Seamless Adaptation of Vision Transformer Fine-tuning for Radar-based Human Activity Recognition
Human Activity Recognition (HAR) such as fall detection has become increasingly critical due to the aging population, necessitating effective monitoring systems to prevent serious injuries and fatalities associated with falls. This study focuses on fine-tuning the Vision Transformer (ViT) model specifically for HAR using radar-based Time-Doppler signatures. Unlike traditional image datasets, these signals present unique challenges due to their non-visual nature and the high degree of similarity among various activities. Directly fine-tuning the ViT with all parameters proves suboptimal for this application. To address this challenge, we propose a novel approach that employs Low-Rank Adaptation (LoRA) fine-tuning in the weight space to facilitate knowledge transfer from pre-trained ViT models. Additionally, to extract fine-grained features, we enhance feature representation through the integration of a serial-parallel adapter in the feature space. Our innovative joint fine-tuning method, tailored for radar-based Time-Doppler signatures, significantly improves HAR accuracy, surpassing existing state-of-the-art methodologies in this domain. Our code is released at https://github.com/wangyijunlyy/SelaFD.
♻ ☆ Crosslingual Capabilities and Knowledge Barriers in Multilingual Large Language Models
Large language models (LLMs) are typically multilingual due to pretraining on diverse multilingual corpora. But can these models relate corresponding concepts across languages, i.e., be crosslingual? This study evaluates state-of-the-art LLMs on inherently crosslingual tasks. We observe that while these models show promising surface-level crosslingual abilities on machine translation and embedding space analyses, they struggle with deeper crosslingual knowledge transfer, revealing a crosslingual knowledge barrier in both general (MMLU benchmark) and domain-specific (Harry Potter quiz and TOFU benchmark) contexts. Since simple inference-time mitigation methods offer only limited improvement, we propose fine-tuning of LLMs on mixed-language data, which effectively reduces these gaps, even when using out-of-domain datasets like WikiText. Our findings suggest the need for explicit optimization to unlock the full crosslingual potential of LLMs. Our code is publicly available at https://github.com/google-research/crosslingual-knowledge-barriers.
♻ ☆ Reflective-Net: Learning from Explanations
We examine whether data generated by explanation techniques, which promote a process of self-reflection, can improve classifier performance. Our work is based on the idea that humans have the ability to make quick, intuitive decisions as well as to reflect on their own thinking and learn from explanations. To the best of our knowledge, this is the first time that the potential of mimicking this process by using explanations generated by explainability methods has been explored. We found that combining explanations with traditional labeled data leads to significant improvements in classification accuracy and training efficiency across multiple image classification datasets and convolutional neural network architectures. It is worth noting that during training, we not only used explanations for the correct or predicted class, but also for other classes. This serves multiple purposes, including allowing for reflection on potential outcomes and enriching the data through augmentation.
Information Retrieval 29
☆ Zero-Shot Complex Question-Answering on Long Scientific Documents AAAI 2025
With the rapid development in Transformer-based language models, the reading comprehension tasks on short documents and simple questions have been largely addressed. Long documents, specifically the scientific documents that are densely packed with knowledge discovered and developed by humans, remain relatively unexplored. These documents often come with a set of complex and more realistic questions, adding to their complexity. We present a zero-shot pipeline framework that enables social science researchers to perform question-answering tasks that are complex yet of predetermined question formats on full-length research papers without requiring machine learning expertise. Our approach integrates pre-trained language models to handle challenging scenarios including multi-span extraction, multi-hop reasoning, and long-answer generation. Evaluating on MLPsych, a novel dataset of social psychology papers with annotated complex questions, we demonstrate that our framework achieves strong performance through combination of extractive and generative models. This work advances document understanding capabilities for social sciences while providing practical tools for researchers.
comment: AAAI 2025 Workshop on Document Understanding and Intelligence
☆ Towards Robust Expert Finding in Community Question Answering Platforms
This paper introduces TUEF, a topic-oriented user-interaction model for fair Expert Finding in Community Question Answering (CQA) platforms. The Expert Finding task in CQA platforms involves identifying proficient users capable of providing accurate answers to questions from the community. To this aim, TUEF improves the robustness and credibility of the CQA platform through a more precise Expert Finding component. The key idea of TUEF is to exploit diverse types of information, specifically, content and social information, to identify more precisely experts thus improving the robustness of the task. We assess TUEF through reproducible experiments conducted on a large-scale dataset from StackOverflow. The results consistently demonstrate that TUEF outperforms state-of-the-art competitors while promoting transparent expert identification.
☆ Are some books better than others?
Scholars, awards committees, and laypeople frequently discuss the merit of written works. Literary professionals and journalists differ in how much perspectivism they concede in their book reviews. Here, we quantify how strongly book reviews are determined by the actual book contents vs. idiosyncratic reader tendencies. In our analysis of 624,320 numerical and textual book reviews, we find that the contents of professionally published books are not predictive of a random reader's reading enjoyment. Online reviews of popular fiction and non-fiction books carry up to ten times more information about the reviewer than about the book. For books of a preferred genre, readers might be less likely to give low ratings, but still struggle to converge in their relative assessments. We find that book evaluations generalize more across experienced review writers than casual readers. When discussing specific issues with a book, one review text had poor predictability of issues brought up in another review of the same book. We conclude that extreme perspectivism is a justifiable position when researching literary quality, bestowing literary awards, and designing recommendation systems.
☆ Personalized Generation In Large Model Era: A Survey
In the era of large models, content generation is gradually shifting to Personalized Generation (PGen), tailoring content to individual preferences and needs. This paper presents the first comprehensive survey on PGen, investigating existing research in this rapidly growing field. We conceptualize PGen from a unified perspective, systematically formalizing its key components, core objectives, and abstract workflows. Based on this unified perspective, we propose a multi-level taxonomy, offering an in-depth review of technical advancements, commonly used datasets, and evaluation metrics across multiple modalities, personalized contexts, and tasks. Moreover, we envision the potential applications of PGen and highlight open challenges and promising directions for future exploration. By bridging PGen research across multiple modalities, this survey serves as a valuable resource for fostering knowledge sharing and interdisciplinary collaboration, ultimately contributing to a more personalized digital landscape.
☆ Efficient Long Sequential Low-rank Adaptive Attention for Click-through rate Prediction
In the context of burgeoning user historical behavior data, Accurate click-through rate(CTR) prediction requires effective modeling of lengthy user behavior sequences. As the volume of such data keeps swelling, the focus of research has shifted towards developing effective long-term behavior modeling methods to capture latent user interests. Nevertheless, the complexity introduced by large scale data brings about computational hurdles. There is a pressing need to strike a balance between achieving high model performance and meeting the strict response time requirements of online services. While existing retrieval-based methods (e.g., similarity filtering or attention approximation) achieve practical runtime efficiency, they inherently compromise information fidelity through aggressive sequence truncation or attention sparsification. This paper presents a novel attention mechanism. It overcomes the shortcomings of existing methods while ensuring computational efficiency. This mechanism learn compressed representation of sequence with length $L$ via low-rank projection matrices (rank $r \ll L$), reducing attention complexity from $O(L)$ to $O(r)$. It also integrates a uniquely designed loss function to preserve nonlinearity of attention. In the inference stage, the mechanism adopts matrix absorption and prestorage strategies. These strategies enable it to effectively satisfy online constraints. Comprehensive offline and online experiments demonstrate that the proposed method outperforms current state-of-the-art solutions.
☆ Sparse Meets Dense: Unified Generative Recommendations with Cascaded Sparse-Dense Representations
Generative models have recently gained attention in recommendation systems by directly predicting item identifiers from user interaction sequences. However, existing methods suffer from significant information loss due to the separation of stages such as quantization and sequence modeling, hindering their ability to achieve the modeling precision and accuracy of sequential dense retrieval techniques. Integrating generative and dense retrieval methods remains a critical challenge. To address this, we introduce the Cascaded Organized Bi-Represented generAtive retrieval (COBRA) framework, which innovatively integrates sparse semantic IDs and dense vectors through a cascading process. Our method alternates between generating these representations by first generating sparse IDs, which serve as conditions to aid in the generation of dense vectors. End-to-end training enables dynamic refinement of dense representations, capturing both semantic insights and collaborative signals from user-item interactions. During inference, COBRA employs a coarse-to-fine strategy, starting with sparse ID generation and refining them into dense vectors via the generative model. We further propose BeamFusion, an innovative approach combining beam search with nearest neighbor scores to enhance inference flexibility and recommendation diversity. Extensive experiments on public datasets and offline tests validate our method's robustness. Online A/B tests on a real-world advertising platform with over 200 million daily users demonstrate substantial improvements in key metrics, highlighting COBRA's practical advantages.
☆ Hierarchical Re-ranker Retriever (HRR)
Retrieving the right level of context for a given query is a perennial challenge in information retrieval - too large a chunk dilutes semantic specificity, while chunks that are too small lack broader context. This paper introduces the Hierarchical Re-ranker Retriever (HRR), a framework designed to achieve both fine-grained and high-level context retrieval for large language model (LLM) applications. In HRR, documents are split into sentence-level and intermediate-level (512 tokens) chunks to maximize vector-search quality for both short and broad queries. We then employ a reranker that operates on these 512-token chunks, ensuring an optimal balance neither too coarse nor too fine for robust relevance scoring. Finally, top-ranked intermediate chunks are mapped to parent chunks (2048 tokens) to provide an LLM with sufficiently large context.
comment: 14 pages
☆ PersonaX: A Recommendation Agent Oriented User Modeling Framework for Long Behavior Sequence
Recommendation agents leverage large language models for user modeling LLM UM to construct textual personas guiding alignment with real users. However existing LLM UM methods struggle with long user generated content UGC due to context limitations and performance degradation. To address this sampling strategies prioritize relevance or recency are often applied yet they inevitably neglect the diverse user interests embedded within the discarded behaviors resulting in incomplete modeling and degraded profiling quality. Furthermore relevance based sampling requires real time retrieval forcing the user modeling process to operate online which introduces significant latency overhead. In this paper we propose PersonaX an agent agnostic LLM UM framework that tackles these challenges through sub behavior sequence SBS selection and offline multi persona construction. PersonaX extracts compact SBS segments offline to capture diverse user interests generating fine grained textual personas that are cached for efficient online retrieval. This approach ensures that the user persona used for prompting remains highly relevant to the current context while eliminating the need for online user modeling. For SBS selection we ensure both efficiency length less than five and high representational quality by balancing prototypicality and diversity within the sampled data. Extensive experiments validate the effectiveness and versatility of PersonaX in high quality user profiling. Utilizing only 30 to 50 percent of the behavioral data with a sequence length of 480 integrating PersonaX with AgentCF yields an absolute performance improvement of 3 to 11 percent while integration with Agent4Rec results in a gain of 10 to 50 percent. PersonaX as an agent agnostic framework sets a new benchmark for scalable user modeling paving the way for more accurate and efficient LLM driven recommendation agents.
comment: draft paper
☆ Towards Explainable Doctor Recommendation with Large Language Models
The advent of internet medicine provides patients with unprecedented convenience in searching and communicating with doctors relevant to their diseases and desired treatments online. However, the current doctor recommendation systems fail to fully ensure the professionalism and interpretability of the recommended results. In this work, we formulate doctor recommendation as a ranking task and develop a large language model (LLM)-based pointwise ranking framework. Our framework ranks doctors according to their relevance regarding specific diseases-treatment pairs in a zero-shot setting. The advantage of our framework lies in its ability to generate precise and explainable doctor ranking results. Additionally, we construct DrRank, a new expertise-driven doctor ranking dataset comprising over 38 disease-treatment pairs. Experiment results on the DrRank dataset demonstrate that our framework significantly outperforms the strongest cross-encoder baseline, achieving a notable gain of +5.45 in the NDCG@10 score while maintaining affordable latency consumption. Furthermore, we comprehensively present the fairness analysis results of our framework from three perspectives of different diseases, patient gender, and geographical regions. Meanwhile, the interpretability of our framework is rigorously verified by three human experts, providing further evidence of the reliability of our proposed framework for doctor recommendation.
comment: 12 pages, 6 figures, Journal of Biomedical and Health Informatics (JBHI) under review
☆ Tailoring Table Retrieval from a Field-aware Hybrid Matching Perspective
Table retrieval, essential for accessing information through tabular data, is less explored compared to text retrieval. The row/column structure and distinct fields of tables (including titles, headers, and cells) present unique challenges. For example, different table fields have varying matching preferences: cells may favor finer-grained (word/phrase level) matching over broader (sentence/passage level) matching due to their fragmented and detailed nature, unlike titles. This necessitates a table-specific retriever to accommodate the various matching needs of each table field. Therefore, we introduce a Table-tailored HYbrid Matching rEtriever (THYME), which approaches table retrieval from a field-aware hybrid matching perspective. Empirical results on two table retrieval benchmarks, NQ-TABLES and OTT-QA, show that THYME significantly outperforms state-of-the-art baselines. Comprehensive analyses confirm the differing matching preferences across table fields and validate the design of THYME.
☆ Semi-Supervised In-Context Learning: A Baseline Study
Most existing work in data selection for In-Context Learning (ICL) has focused on constructing demonstrations from ground truth annotations, with limited attention given to selecting reliable self-generated annotations. In this work, we propose a three-step semi-supervised ICL framework: annotation generation, demonstration selection, and semi-supervised inference. Our baseline, Naive-SemiICL, which prompts select high-confidence self-generated demonstrations for ICL prompting, outperforms a 16-shot baseline by an average of 9.94% across 16 datasets. We further introduce IterPSD, an annotation approach that refines pseudo-demonstrations iteratively, achieving up to 6.8% additional gains in classification tasks. Lastly, we reveal a scaling law for semi-supervised ICL, where models achieve optimal performance with over 1,000 demonstrations.
☆ Zero-Shot Multi-Label Classification of Bangla Documents: Large Decoders Vs. Classic Encoders
Bangla, a language spoken by over 300 million native speakers and ranked as the sixth most spoken language worldwide, presents unique challenges in natural language processing (NLP) due to its complex morphological characteristics and limited resources. While recent Large Decoder Based models (LLMs), such as GPT, LLaMA, and DeepSeek, have demonstrated excellent performance across many NLP tasks, their effectiveness in Bangla remains largely unexplored. In this paper, we establish the first benchmark comparing decoder-based LLMs with classic encoder-based models for Zero-Shot Multi-Label Classification (Zero-Shot-MLC) task in Bangla. Our evaluation of 32 state-of-the-art models reveals that, existing so-called powerful encoders and decoders still struggle to achieve high accuracy on the Bangla Zero-Shot-MLC task, suggesting a need for more research and resources for Bangla NLP.
☆ ExpertGenQA: Open-ended QA generation in Specialized Domains
Generating high-quality question-answer pairs for specialized technical domains remains challenging, with existing approaches facing a tradeoff between leveraging expert examples and achieving topical diversity. We present ExpertGenQA, a protocol that combines few-shot learning with structured topic and style categorization to generate comprehensive domain-specific QA pairs. Using U.S. Federal Railroad Administration documents as a test bed, we demonstrate that ExpertGenQA achieves twice the efficiency of baseline few-shot approaches while maintaining $94.4\%$ topic coverage. Through systematic evaluation, we show that current LLM-based judges and reward models exhibit strong bias toward superficial writing styles rather than content quality. Our analysis using Bloom's Taxonomy reveals that ExpertGenQA better preserves the cognitive complexity distribution of expert-written questions compared to template-based approaches. When used to train retrieval models, our generated queries improve top-1 accuracy by $13.02\%$ over baseline performance, demonstrating their effectiveness for downstream applications in technical domains.
☆ Optimizing open-domain question answering with graph-based retrieval augmented generation
In this work, we benchmark various graph-based retrieval-augmented generation (RAG) systems across a broad spectrum of query types, including OLTP-style (fact-based) and OLAP-style (thematic) queries, to address the complex demands of open-domain question answering (QA). Traditional RAG methods often fall short in handling nuanced, multi-document synthesis tasks. By structuring knowledge as graphs, we can facilitate the retrieval of context that captures greater semantic depth and enhances language model operations. We explore graph-based RAG methodologies and introduce TREX, a novel, cost-effective alternative that combines graph-based and vector-based retrieval techniques. Our benchmarking across four diverse datasets highlights the strengths of different RAG methodologies, demonstrates TREX's ability to handle multiple open-domain QA types, and reveals the limitations of current evaluation methods. In a real-world technical support case study, we demonstrate how TREX solutions can surpass conventional vector-based RAG in efficiently synthesizing data from heterogeneous sources. Our findings underscore the potential of augmenting large language models with advanced retrieval and orchestration capabilities, advancing scalable, graph-based AI solutions.
♻ ☆ MAPS: Motivation-Aware Personalized Search via LLM-Driven Consultation Alignment
Personalized product search aims to retrieve and rank items that match users' preferences and search intent. Despite their effectiveness, existing approaches typically assume that users' query fully captures their real motivation. However, our analysis of a real-world e-commerce platform reveals that users often engage in relevant consultations before searching, indicating they refine intents through consultations based on motivation and need. The implied motivation in consultations is a key enhancing factor for personalized search. This unexplored area comes with new challenges including aligning contextual motivations with concise queries, bridging the category-text gap, and filtering noise within sequence history. To address these, we propose a Motivation-Aware Personalized Search (MAPS) method. It embeds queries and consultations into a unified semantic space via LLMs, utilizes a Mixture of Attention Experts (MoAE) to prioritize critical semantics, and introduces dual alignment: (1) contrastive learning aligns consultations, reviews, and product features; (2) bidirectional attention integrates motivation-aware embeddings with user preferences. Extensive experiments on real and synthetic data show MAPS outperforms existing methods in both retrieval and ranking tasks.
♻ ☆ A Comparative Evaluation of Quantification Methods
Quantification represents the problem of estimating the distribution of class labels on unseen data. It also represents a growing research field in supervised machine learning, for which a large variety of different algorithms has been proposed in recent years. However, a comprehensive empirical comparison of quantification methods that supports algorithm selection is not available yet. In this work, we close this research gap by conducting a thorough empirical performance comparison of 24 different quantification methods on overall more than 40 data sets, considering binary as well as multiclass quantification settings. We observe that no single algorithm generally outperforms all competitors, but identify a group of methods including the threshold selection-based Median Sweep and TSMax methods, the DyS framework including the HDy method, Forman's mixture model, and Friedman's method that performs best in the binary setting. For the multiclass setting, we observe that a different, broad group of algorithms yields good performance, including the HDx method, the Generalized Probabilistic Adjusted Count, the readme method, the energy distance minimization method, the EM algorithm for quantification, and Friedman's method. We also find that tuning the underlying classifiers has in most cases only a limited impact on the quantification performance. More generally, we find that the performance on multiclass quantification is inferior to the results obtained in the binary setting. Our results can guide practitioners who intend to apply quantification algorithms and help researchers to identify opportunities for future research.
comment: 41 pages, 18 figures, 9 tables
♻ ☆ A Comprehensive Survey on Composed Image Retrieval
Composed Image Retrieval (CIR) is an emerging yet challenging task that allows users to search for target images using a multimodal query, comprising a reference image and a modification text specifying the user's desired changes to the reference image. Given its significant academic and practical value, CIR has become a rapidly growing area of interest in the computer vision and machine learning communities, particularly with the advances in deep learning. To the best of our knowledge, there is currently no comprehensive review of CIR to provide a timely overview of this field. Therefore, we synthesize insights from over 120 publications in top conferences and journals, including ACM TOIS, SIGIR, and CVPR In particular, we systematically categorize existing supervised CIR and zero-shot CIR models using a fine-grained taxonomy. For a comprehensive review, we also briefly discuss approaches for tasks closely related to CIR, such as attribute-based CIR and dialog-based CIR. Additionally, we summarize benchmark datasets for evaluation and analyze existing supervised and zero-shot CIR methods by comparing experimental results across multiple datasets. Furthermore, we present promising future directions in this field, offering practical insights for researchers interested in further exploration. The curated collection of related works is maintained and continuously updated in https://github.com/haokunwen/Awesome-Composed-Image-Retrieval.
♻ ☆ GENET: Unleashing the Power of Side Information for Recommendation via Hypergraph Pre-training
Recommendation with side information has drawn significant research interest due to its potential to mitigate user feedback sparsity. However, existing models struggle with generalization across diverse domains and types of side information. In particular, three challenges have not been addressed, and they are (1) the diverse formats of side information, including text sequences. (2) The diverse semantics of side information that describes items and users from multi-level in a context different from recommendation systems. (3) The diverse correlations in side information to measure similarity over multiple objects beyond pairwise relations. In this paper, we introduce GENET (Generalized hypErgraph pretraiNing on sidE informaTion), which pre-trains user and item representations on feedback-irrelevant side information and fine-tunes the representations on user feedback data. GENET leverages pre-training as a means to prevent side information from overshadowing critical ID features and feedback signals. It employs a hypergraph framework to accommodate various types of diverse side information. During pre-training, GENET integrates tasks for hyperlink prediction and self-supervised contrast to capture fine-grained semantics at both local and global levels. Additionally, it introduces a unique strategy to enhance pre-training robustness by perturbing positive samples while maintaining high-order relations. Extensive experiments demonstrate that GENET exhibits strong generalization capabilities, outperforming the SOTA method by up to 38% in TOP-N recommendation and Sequential recommendation tasks on various datasets with different side information.
♻ ☆ From Matching to Generation: A Survey on Generative Information Retrieval
Information Retrieval (IR) systems are crucial tools for users to access information, which have long been dominated by traditional methods relying on similarity matching. With the advancement of pre-trained language models, generative information retrieval (GenIR) emerges as a novel paradigm, attracting increasing attention. Based on the form of information provided to users, current research in GenIR can be categorized into two aspects: \textbf{(1) Generative Document Retrieval} (GR) leverages the generative model's parameters for memorizing documents, enabling retrieval by directly generating relevant document identifiers without explicit indexing. \textbf{(2) Reliable Response Generation} employs language models to directly generate information users seek, breaking the limitations of traditional IR in terms of document granularity and relevance matching while offering flexibility, efficiency, and creativity to meet practical needs. This paper aims to systematically review the latest research progress in GenIR. We will summarize the advancements in GR regarding model training and structure, document identifier, incremental learning, etc., as well as progress in reliable response generation in aspects of internal knowledge memorization, external knowledge augmentation, etc. We also review the evaluation, challenges and future developments in GenIR systems. This review aims to offer a comprehensive reference for researchers, encouraging further development in the GenIR field. Github Repository: https://github.com/RUC-NLPIR/GenIR-Survey
♻ ☆ Pattern-wise Transparent Sequential Recommendation
A transparent decision-making process is essential for developing reliable and trustworthy recommender systems. For sequential recommendation, it means that the model can identify key items that account for its recommendation results. However, achieving both interpretability and recommendation performance simultaneously is challenging, especially for models that take the entire sequence of items as input without screening. In this paper, we propose an interpretable framework (named PTSR) that enables a pattern-wise transparent decision-making process without extra features. It breaks the sequence of items into multi-level patterns that serve as atomic units throughout the recommendation process. The contribution of each pattern to the outcome is quantified in the probability space. With a carefully designed score correction mechanism, the pattern contribution can be implicitly learned in the absence of ground-truth key patterns. The final recommended items are those that most key patterns strongly endorse. Extensive experiments on five public datasets demonstrate remarkable recommendation performance, while statistical analysis and case studies validate the model interpretability.
comment: This paper has been accepted by IEEE TKDE
♻ ☆ VideoRAG: Retrieval-Augmented Generation over Video Corpus
Retrieval-Augmented Generation (RAG) is a powerful strategy for improving the factual accuracy of models by retrieving external knowledge relevant to queries and incorporating it into the generation process. However, existing approaches primarily focus on text, with some recent advancements considering images, and they largely overlook videos, a rich source of multimodal knowledge capable of representing contextual details more effectively than any other modality. While very recent studies explore the use of videos in response generation, they either predefine query-associated videos without retrieval or convert videos into textual descriptions losing multimodal richness. To tackle these, we introduce VideoRAG, a framework that not only dynamically retrieves videos based on their relevance with queries but also utilizes both visual and textual information. The operation of VideoRAG is powered by recent Large Video Language Models (LVLMs), which enable the direct processing of video content to represent it for retrieval and the seamless integration of retrieved videos jointly with queries for response generation. Also, inspired by that the context size of LVLMs may not be sufficient to process all frames in extremely long videos and not all frames are equally important, we introduce a video frame selection mechanism to extract the most informative subset of frames, along with a strategy to extract textual information from videos (as it can aid the understanding of video content) when their subtitles are not available. We experimentally validate the effectiveness of VideoRAG, showcasing that it is superior to relevant baselines. Code is available at https://github.com/starsuzi/VideoRAG.
♻ ☆ Scholar Name Disambiguation with Search-enhanced LLM Across Language
The task of scholar name disambiguation is crucial in various real-world scenarios, including bibliometric-based candidate evaluation for awards, application material anti-fraud measures, and more. Despite significant advancements, current methods face limitations due to the complexity of heterogeneous data, often necessitating extensive human intervention. This paper proposes a novel approach by leveraging search-enhanced language models across multiple languages to improve name disambiguation. By utilizing the powerful query rewriting, intent recognition, and data indexing capabilities of search engines, our method can gather richer information for distinguishing between entities and extracting profiles, resulting in a more comprehensive data dimension. Given the strong cross-language capabilities of large language models(LLMs), optimizing enhanced retrieval methods with this technology offers substantial potential for high-efficiency information retrieval and utilization. Our experiments demonstrate that incorporating local languages significantly enhances disambiguation performance, particularly for scholars from diverse geographic regions. This multi-lingual, search-enhanced methodology offers a promising direction for more efficient and accurate active scholar name disambiguation.
♻ ☆ Hierarchical Causal Transformer with Heterogeneous Information for Expandable Sequential Recommendation
Sequential recommendation systems leveraging transformer architectures have demonstrated exceptional capabilities in capturing user behavior patterns. At the core of these systems lies the critical challenge of constructing effective item representations. Traditional approaches employ feature fusion through simple concatenation or basic neural architectures to create uniform representation sequences. However, these conventional methods fail to address the intrinsic diversity of item attributes, thereby constraining the transformer's capacity to discern fine-grained patterns and hindering model extensibility. Although recent research has begun incorporating user-related heterogeneous features into item sequences, the equally crucial item-side heterogeneous feature continue to be neglected. To bridge this methodological gap, we present HeterRec - an innovative framework featuring two novel components: the Heterogeneous Token Flattening Layer (HTFL) and Hierarchical Causal Transformer (HCT). HTFL pioneers a sophisticated tokenization mechanism that decomposes items into multi-dimensional token sets and structures them into heterogeneous sequences, enabling scalable performance enhancement through model expansion. The HCT architecture further enhances pattern discovery through token-level and item-level attention mechanisms. furthermore, we develop a Listwise Multi-step Prediction (LMP) objective function to optimize learning process. Rigorous validation, including real-world industrial platforms, confirms HeterRec's state-of-the-art performance in both effective and efficiency.
♻ ☆ Accuracy Assessment of OpenAlex and Clarivate Scholar ID with an LLM-Assisted Benchmark
In quantitative SciSci (science of science) studies, accurately identifying individual scholars is paramount for scientific data analysis. However, the variability in how names are represented-due to commonality, abbreviations, and different spelling conventions-complicates this task. While identifier systems like ORCID are being developed, many scholars remain unregistered, and numerous publications are not included. Scholarly databases such as Clarivate and OpenAlex have introduced their own ID systems as preliminary name disambiguation solutions. This study evaluates the effectiveness of these systems across different groups to determine their suitability for various application scenarios. We sampled authors from the top quartile (Q1) of Web of Science (WOS) journals based on country, discipline, and number of corresponding author papers. For each group, we selected 100 scholars and meticulously annotated all their papers using a Search-enhanced Large Language Model method. Using these annotations, we identified the corresponding IDs in OpenAlex and Clarivate, extracted all associated papers, filtered for Q1 WOS journals, and calculated precision and recall by comparing against the annotated dataset.
♻ ☆ Hashtag Re-Appropriation for Audience Control on Recommendation-Driven Social Media Xiaohongshu (rednote)
Algorithms have played a central role in personalized recommendations on social media. However, they also present significant obstacles for content creators trying to predict and manage their audience reach. This issue is particularly challenging for marginalized groups seeking to maintain safe spaces. Our study explores how women on Xiaohongshu (rednote), a recommendation-driven social platform, proactively re-appropriate hashtags (e.g., #Baby Supplemental Food) by using them in posts unrelated to their literal meaning. The hashtags were strategically chosen from topics that would be uninteresting to the male audience they wanted to block. Through a mixed-methods approach, we analyzed the practice of hashtag re-appropriation based on 5,800 collected posts and interviewed 24 active users from diverse backgrounds to uncover users' motivations and reactions towards the re-appropriation. This practice highlights how users can reclaim agency over content distribution on recommendation-driven platforms, offering insights into self-governance within algorithmic-centered power structures.
♻ ☆ Lightweight yet Efficient: An External Attentive Graph Convolutional Network with Positional Prompts for Sequential Recommendation
Graph-based Sequential Recommender systems (GSRs) have gained significant research attention due to their ability to simultaneously handle user-item interactions and sequential relationships between items. Current GSRs often utilize composite or in-depth structures for graph encoding (e.g., the Graph Transformer). Nevertheless, they have high computational complexity, hindering the deployment on resource-constrained edge devices. Moreover, the relative position encoding in Graph Transformer has difficulty in considering the complicated positional dependencies within sequence. To this end, we propose an External Attentive Graph convolutional network with Positional prompts for Sequential recommendation, namely EA-GPS. Specifically, we first introduce an external attentive graph convolutional network that linearly measures the global associations among nodes via two external memory units. Then, we present a positional prompt-based decoder that explicitly treats the absolute item positions as external prompts. By introducing length-adaptive sequential masking and a soft attention network, such a decoder facilitates the model to capture the long-term positional dependencies and contextual relationships within sequences. Extensive experimental results on five real-world datasets demonstrate that the proposed EA-GPS outperforms the state-of-the-art methods. Remarkably, it achieves the superior performance while maintaining a smaller parameter size and lower training overhead. The implementation of this work is publicly available at https://github.com/ZZY-GraphMiningLab/EA-GPS.
comment: 26 pages, 8 figures, journal paper, accepted by TOIS at 20th February, 2025
♻ ☆ GaVaMoE: Gaussian-Variational Gated Mixture of Experts for Explainable Recommendation
Large language model-based explainable recommendation (LLM-based ER) systems show promise in generating human-like explanations for recommendations. However, they face challenges in modeling user-item collaborative preferences, personalizing explanations, and handling sparse user-item interactions. To address these issues, we propose GaVaMoE, a novel Gaussian-Variational Gated Mixture of Experts framework for explainable recommendation. GaVaMoE introduces two key components: (1) a rating reconstruction module that employs Variational Autoencoder (VAE) with a Gaussian Mixture Model (GMM) to capture complex user-item collaborative preferences, serving as a pre-trained multi-gating mechanism; and (2) a set of fine-grained expert models coupled with the multi-gating mechanism for generating highly personalized explanations. The VAE component models latent factors in user-item interactions, while the GMM clusters users with similar behaviors. Each cluster corresponds to a gate in the multi-gating mechanism, routing user-item pairs to appropriate expert models. This architecture enables GaVaMoE to generate tailored explanations for specific user types and preferences, mitigating data sparsity by leveraging user similarities. Extensive experiments on three real-world datasets demonstrate that GaVaMoE significantly outperforms existing methods in explanation quality, personalization, and consistency. Notably, GaVaMoE exhibits robust performance in scenarios with sparse user-item interactions, maintaining high-quality explanations even for users with limited historical data.
♻ ☆ CoRNStack: High-Quality Contrastive Data for Better Code Retrieval and Reranking ICLR 2025
Effective code retrieval plays a crucial role in advancing code generation, bug fixing, and software maintenance, particularly as software systems increase in complexity. While current code embedding models have demonstrated promise in retrieving code snippets for small-scale, well-defined tasks, they often underperform in more demanding real-world applications such as bug localization within GitHub repositories. We hypothesize that a key issue is their reliance on noisy and inconsistent datasets for training, which impedes their ability to generalize to more complex retrieval scenarios. To address these limitations, we introduce CoRNStack, a large-scale, high-quality contrastive training dataset for code that spans multiple programming languages. This dataset is curated using consistency filtering to eliminate noisy positives and is further enriched with mined hard negatives, thereby facilitating more effective learning. We demonstrate that contrastive training of embedding models using CoRNStack leads to state-of-the-art performance across a variety of code retrieval tasks. Furthermore, the dataset can be leveraged for training code reranking models, a largely underexplored area compared to text reranking. Our finetuned code reranking model significantly improves the ranking quality over the retrieved results. Finally, by employing our code retriever and reranker together, we demonstrate significant improvements in function localization for GitHub issues, an important component of real-world software development.
comment: Published as a conference paper at ICLR 2025. First and second author had equal contribution
♻ ☆ A Simplified Retriever to Improve Accuracy of Phenotype Normalizations by Large Language Models
Large language models (LLMs) have shown improved accuracy in phenotype term normalization tasks when augmented with retrievers that suggest candidate normalizations based on term definitions. In this work, we introduce a simplified retriever that enhances LLM accuracy by searching the Human Phenotype Ontology (HPO) for candidate matches using contextual word embeddings from BioBERT without the need for explicit term definitions. Testing this method on terms derived from the clinical synopses of Online Mendelian Inheritance in Man (OMIM), we demonstrate that the normalization accuracy of a state-of-the-art LLM increases from a baseline of 62.3% without augmentation to 90.3% with retriever augmentation. This approach is potentially generalizable to other biomedical term normalization tasks and offers an efficient alternative to more complex retrieval methods.
comment: Published by Frontiers in Digital Health
Computation and Language 81
☆ Forgetting Transformer: Softmax Attention with a Forget Gate ICLR 2025
An essential component of modern recurrent sequence models is the forget gate. While Transformers do not have an explicit recurrent form, we show that a forget gate can be naturally incorporated into Transformers by down-weighting the unnormalized attention scores in a data-dependent way. We name this attention mechanism the Forgetting Attention and the resulting model the Forgetting Transformer (FoX). We show that FoX outperforms the Transformer on long-context language modeling, length extrapolation, and short-context downstream tasks, while performing on par with the Transformer on long-context downstream tasks. Moreover, it is compatible with the FlashAttention algorithm and does not require any positional embeddings. Several analyses, including the needle-in-the-haystack test, show that FoX also retains the Transformer's superior long-context capabilities over recurrent sequence models such as Mamba-2, HGRN2, and DeltaNet. We also introduce a "Pro" block design that incorporates some common architectural components in recurrent sequence models and find it significantly improves the performance of both FoX and the Transformer. Our code is available at https://github.com/zhixuan-lin/forgetting-transformer.
comment: Published as a conference paper at ICLR 2025
☆ Superficial Self-Improved Reasoners Benefit from Model Merging
As scaled language models (LMs) approach human-level reasoning capabilities, self-improvement emerges as a solution to synthesizing high-quality data corpus. While previous research has identified model collapse as a risk in self-improvement, where model outputs become increasingly deterministic, we discover a more fundamental challenge: the superficial self-improved reasoners phenomenon. In particular, our analysis reveals that even when LMs show improved in-domain (ID) reasoning accuracy, they actually compromise their generalized reasoning capabilities on out-of-domain (OOD) tasks due to memorization rather than genuine. Through a systematic investigation of LM architecture, we discover that during self-improvement, LM weight updates are concentrated in less reasoning-critical layers, leading to superficial learning. To address this, we propose Iterative Model Merging (IMM), a method that strategically combines weights from original and self-improved models to preserve generalization while incorporating genuine reasoning improvements. Our approach effectively mitigates both LM collapse and superficial learning, moving towards more stable self-improving systems.
☆ Provable Benefits of Task-Specific Prompts for In-context Learning AISTATS
The in-context learning capabilities of modern language models have motivated a deeper mathematical understanding of sequence models. A line of recent work has shown that linear attention models can emulate projected gradient descent iterations to implicitly learn the task vector from the data provided in the context window. In this work, we consider a novel setting where the global task distribution can be partitioned into a union of conditional task distributions. We then examine the use of task-specific prompts and prediction heads for learning the prior information associated with the conditional task distribution using a one-layer attention model. Our results on loss landscape show that task-specific prompts facilitate a covariance-mean decoupling where prompt-tuning explains the conditional mean of the distribution whereas the variance is learned/explained through in-context learning. Incorporating task-specific head further aids this process by entirely decoupling estimation of mean and variance components. This covariance-mean perspective similarly explains how jointly training prompt and attention weights can provably help over fine-tuning after pretraining.
comment: Proceedings of the 28th International Conference on Artificial Intelligence and Statistics (AISTATS) 2025
☆ Twenty Years of Personality Computing: Threats, Challenges and Future Directions
Personality Computing is a field at the intersection of Personality Psychology and Computer Science. Started in 2005, research in the field utilizes computational methods to understand and predict human personality traits. The expansion of the field has been very rapid and, by analyzing digital footprints (text, images, social media, etc.), it helped to develop systems that recognize and even replicate human personality. While offering promising applications in talent recruiting, marketing and healthcare, the ethical implications of Personality Computing are significant. Concerns include data privacy, algorithmic bias, and the potential for manipulation by personality-aware Artificial Intelligence. This paper provides an overview of the field, explores key methodologies, discusses the challenges and threats, and outlines potential future directions for responsible development and deployment of Personality Computing technologies.
☆ Linear Representations of Political Perspective Emerge in Large Language Models ICLR 2025
Large language models (LLMs) have demonstrated the ability to generate text that realistically reflects a range of different subjective human perspectives. This paper studies how LLMs are seemingly able to reflect more liberal versus more conservative viewpoints among other political perspectives in American politics. We show that LLMs possess linear representations of political perspectives within activation space, wherein more similar perspectives are represented closer together. To do so, we probe the attention heads across the layers of three open transformer-based LLMs (\texttt{Llama-2-7b-chat}, \texttt{Mistral-7b-instruct}, \texttt{Vicuna-7b}). We first prompt models to generate text from the perspectives of different U.S.~lawmakers. We then identify sets of attention heads whose activations linearly predict those lawmakers' DW-NOMINATE scores, a widely-used and validated measure of political ideology. We find that highly predictive heads are primarily located in the middle layers, often speculated to encode high-level concepts and tasks. Using probes only trained to predict lawmakers' ideology, we then show that the same probes can predict measures of news outlets' slant from the activations of models prompted to simulate text from those news outlets. These linear probes allow us to visualize, interpret, and monitor ideological stances implicitly adopted by an LLM as it generates open-ended responses. Finally, we demonstrate that by applying linear interventions to these attention heads, we can steer the model outputs toward a more liberal or conservative stance. Overall, our research suggests that LLMs possess a high-level linear representation of American political ideology and that by leveraging recent advances in mechanistic interpretability, we can identify, monitor, and steer the subjective perspective underlying generated text.
comment: Published as a conference paper at ICLR 2025
☆ Superscopes: Amplifying Internal Feature Representations for Language Model Interpretation
Understanding and interpreting the internal representations of large language models (LLMs) remains an open challenge. Patchscopes introduced a method for probing internal activations by patching them into new prompts, prompting models to self-explain their hidden representations. We introduce Superscopes, a technique that systematically amplifies superposed features in MLP outputs (multilayer perceptron) and hidden states before patching them into new contexts. Inspired by the "features as directions" perspective and the Classifier-Free Guidance (CFG) approach from diffusion models, Superscopes amplifies weak but meaningful features, enabling the interpretation of internal representations that previous methods failed to explain-all without requiring additional training. This approach provides new insights into how LLMs build context and represent complex concepts, further advancing mechanistic interpretability.
☆ Hebbian learning the local structure of language
Learning in the brain is local and unsupervised (Hebbian). We derive the foundations of an effective human language model inspired by these microscopic constraints. It has two parts: (1) a hierarchy of neurons which learns to tokenize words from text (whichiswhatyoudowhenyoureadthis); and (2) additional neurons which bind the learned symanticless patterns of the tokenizer into a symanticful token (an embedding). The model permits continuous parallel learning without forgetting; and is a powerful tokenizer which performs renormalization group. This allows it to exploit redundancy, such that it generates tokens which are always decomposable into a basis set (e.g an alphabet), and can mix features learned from multiple languages. We find that the structure of this model allows it to learn a natural language morphology WITHOUT data. The language data generated by this model predicts the correct distribution of word-forming patterns observed in real languages, and further demonstrates why microscopically human speech is broken up into words. This model provides the basis for understanding the microscopic origins of language and human creativity.
comment: 10 figures, 14 pages
☆ CareerBERT: Matching Resumes to ESCO Jobs in a Shared Embedding Space for Generic Job Recommendations
The rapidly evolving labor market, driven by technological advancements and economic shifts, presents significant challenges for traditional job matching and consultation services. In response, we introduce an advanced support tool for career counselors and job seekers based on CareerBERT, a novel approach that leverages the power of unstructured textual data sources, such as resumes, to provide more accurate and comprehensive job recommendations. In contrast to previous approaches that primarily focus on job recommendations based on a fixed set of concrete job advertisements, our approach involves the creation of a corpus that combines data from the European Skills, Competences, and Occupations (ESCO) taxonomy and EURopean Employment Services (EURES) job advertisements, ensuring an up-to-date and well-defined representation of general job titles in the labor market. Our two-step evaluation approach, consisting of an application-grounded evaluation using EURES job advertisements and a human-grounded evaluation using real-world resumes and Human Resources (HR) expert feedback, provides a comprehensive assessment of CareerBERT's performance. Our experimental results demonstrate that CareerBERT outperforms both traditional and state-of-the-art embedding approaches while showing robust effectiveness in human expert evaluations. These results confirm the effectiveness of CareerBERT in supporting career consultants by generating relevant job recommendations based on resumes, ultimately enhancing the efficiency of job consultations and expanding the perspectives of job seekers. This research contributes to the field of NLP and job recommendation systems, offering valuable insights for both researchers and practitioners in the domain of career consulting and job matching.
comment: Accepted at Expert Systems with Applications. In Press, see https://doi.org/10.1016/j.eswa.2025.127043
☆ EPEE: Towards Efficient and Effective Foundation Models in Biomedicine
Foundation models, including language models, e.g., GPT, and vision models, e.g., CLIP, have significantly advanced numerous biomedical tasks. Despite these advancements, the high inference latency and the "overthinking" issues in model inference impair the efficiency and effectiveness of foundation models, thus limiting their application in real-time clinical settings. To address these challenges, we proposed EPEE (Entropy- and Patience-based Early Exiting), a novel hybrid strategy designed to improve the inference efficiency of foundation models. The core idea was to leverage the strengths of entropy-based and patience-based early exiting methods to overcome their respective weaknesses. To evaluate EPEE, we conducted experiments on three core biomedical tasks-classification, relation extraction, and event extraction-using four foundation models (BERT, ALBERT, GPT-2, and ViT) across twelve datasets, including clinical notes and medical images. The results showed that EPEE significantly reduced inference time while maintaining or improving accuracy, demonstrating its adaptability to diverse datasets and tasks. EPEE addressed critical barriers to deploying foundation models in healthcare by balancing efficiency and effectiveness. It potentially provided a practical solution for real-time clinical decision-making with foundation models, supporting reliable and efficient workflows.
comment: Submitted to npj Digital Medicine
♻ ☆ Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?
The advent of test-time scaling in large language models (LLMs), exemplified by OpenAI's o1 series, has advanced reasoning capabilities by scaling computational resource allocation during inference. While successors like QwQ, Deepseek-R1 (R1) and LIMO replicate these advancements, whether these models truly possess test-time scaling capabilities remains underexplored. This study found that longer CoTs of these o1-like models do not consistently enhance accuracy; in fact, correct solutions are often shorter than incorrect ones for the same questions. Further investigation shows this phenomenon is closely related to models' self-revision capabilities - longer CoTs contain more self-revisions, which often lead to performance degradation. We then compare sequential and parallel scaling strategies on QwQ, R1 and LIMO, finding that parallel scaling achieves better coverage and scalability. Based on these insights, we propose Shortest Majority Vote, a method that combines parallel scaling strategies with CoT length characteristics, significantly improving models' test-time scalability compared to conventional majority voting approaches.
comment: Add the github link
♻ ☆ CNsum:Automatic Summarization for Chinese News Text
Obtaining valuable information from massive data efficiently has become our research goal in the era of Big Data. Text summarization technology has been continuously developed to meet this demand. Recent work has also shown that transformer-based pre-trained language models have achieved great success on various tasks in Natural Language Processing (NLP). Aiming at the problem of Chinese news text summary generation and the application of Transformer structure on Chinese, this paper proposes a Chinese news text summarization model (CNsum) based on Transformer structure, and tests it on Chinese datasets such as THUCNews. The results of the conducted experiments show that CNsum achieves better ROUGE score than the baseline models, which verifies the outperformance of the model.
comment: This withdrawal is due to the lack of authorization from all co-authors for the publication of this version
♻ ☆ Evaluating Intelligence via Trial and Error
Intelligence is a crucial trait for species to find solutions within a limited number of trial-and-error attempts. Building on this idea, we introduce Survival Game as a framework to evaluate intelligence based on the number of failed attempts in a trial-and-error process. Fewer failures indicate higher intelligence. When the expectation and variance of failure counts are both finite, it signals the ability to consistently find solutions to new challenges, which we define as the Autonomous Level of intelligence. Using Survival Game, we comprehensively evaluate existing AI systems. Our results show that while AI systems achieve the Autonomous Level in simple tasks, they are still far from it in more complex tasks, such as vision, search, recommendation, and language. While scaling current AI technologies might help, this would come at an astronomical cost. Projections suggest that achieving the Autonomous Level for general tasks would require $10^{26}$ parameters. To put this into perspective, loading such a massive model requires so many H100 GPUs that their total value is $10^{7}$ times that of Apple Inc.'s market value. Even with Moore's Law, supporting such a parameter scale would take $70$ years. This staggering cost highlights the complexity of human tasks and the inadequacies of current AI technologies. To further investigate this phenomenon, we conduct a theoretical analysis of Survival Game and its experimental results. Our findings suggest that human tasks possess a criticality property. As a result, Autonomous Level requires a deep understanding of the task's underlying mechanisms. Current AI systems, however, do not fully grasp these mechanisms and instead rely on superficial mimicry, making it difficult for them to reach an autonomous level. We believe Survival Game can not only guide the future development of AI but also offer profound insights into human intelligence.
♻ ☆ Optimize Incompatible Parameters through Compatibility-aware Knowledge Integration AAAI'25
Deep neural networks have become foundational to advancements in multiple domains, including recommendation systems, natural language processing, and so on. Despite their successes, these models often contain incompatible parameters that can be underutilized or detrimental to model performance, particularly when faced with specific, varying data distributions. Existing research excels in removing such parameters or merging the outputs of multiple different pretrained models. However, the former focuses on efficiency rather than performance, while the latter requires several times more computing and storage resources to support inference. In this paper, we set the goal to explicitly improve these incompatible parameters by leveraging the complementary strengths of different models, thereby directly enhancing the models without any additional parameters. Specifically, we propose Compatibility-aware Knowledge Integration (CKI), which consists of Parameter Compatibility Assessment and Parameter Splicing, which are used to evaluate the knowledge content of multiple models and integrate the knowledge into one model, respectively. The integrated model can be used directly for inference or for further fine-tuning. We conduct extensive experiments on various datasets for recommendation and language tasks, and the results show that Compatibility-aware Knowledge Integration can effectively optimize incompatible parameters under multiple tasks and settings to break through the training limit of the original model without increasing the inference cost.
comment: Published on AAAI'25(Oral): The Annual AAAI Conference on Artificial Intelligence
♻ ☆ The ShareLM Collection and Plugin: Contributing Human-Model Chats for the Benefit of the Community
Human-model conversations provide a window into users' real-world scenarios, behavior, and needs, and thus are a valuable resource for model development and research. While for-profit companies collect user data through the APIs of their models, using it internally to improve their own models, the open source and research community lags behind. We introduce the ShareLM collection, a unified set of human conversations with large language models, and its accompanying plugin, a Web extension for voluntarily contributing user-model conversations. Where few platforms share their chats, the ShareLM plugin adds this functionality, thus, allowing users to share conversations from most platforms. The plugin allows the user to rate their conversations, both at the conversation and the response levels, and delete conversations they prefer to keep private before they ever leave the user's local storage. We release the plugin conversations as part of the ShareLM collection, and call for more community effort in the field of open human-model data. The code, plugin, and data are available.
♻ ☆ MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypotheses ICLR 2025
Scientific discovery contributes largely to human society's prosperity, and recent progress shows that LLMs could potentially catalyze this process. However, it is still unclear whether LLMs can discover novel and valid hypotheses in chemistry. In this work, we investigate this central research question: Can LLMs automatically discover novel and valid chemistry research hypotheses given only a chemistry research background (consisting of a research question and/or a background survey), without limitation on the domain of the research question? After extensive discussions with chemistry experts, we propose an assumption that a majority of chemistry hypotheses can be resulted from a research background and several inspirations. With this key insight, we break the central question into three smaller fundamental questions. In brief, they are: (1) given a background question, whether LLMs can retrieve good inspirations; (2) with background and inspirations, whether LLMs can lead to hypothesis; and (3) whether LLMs can identify good hypotheses to rank them higher. To investigate these questions, we construct a benchmark consisting of 51 chemistry papers published in Nature, Science, or a similar level in 2024 (all papers are only available online since 2024). Every paper is divided by chemistry PhD students into three components: background, inspirations, and hypothesis. The goal is to rediscover the hypothesis, given only the background and a large randomly selected chemistry literature corpus consisting the ground truth inspiration papers, with LLMs trained with data up to 2023. We also develop an LLM-based multi-agent framework that leverages the assumption, consisting of three stages reflecting the three smaller questions. The proposed method can rediscover many hypotheses with very high similarity with the ground truth ones, covering the main innovations.
comment: Accepted by ICLR 2025
♻ ☆ NavRAG: Generating User Demand Instructions for Embodied Navigation through Retrieval-Augmented LLM
Vision-and-Language Navigation (VLN) is an essential skill for embodied agents, allowing them to navigate in 3D environments following natural language instructions. High-performance navigation models require a large amount of training data, the high cost of manually annotating data has seriously hindered this field. Therefore, some previous methods translate trajectory videos into step-by-step instructions for expanding data, but such instructions do not match well with users' communication styles that briefly describe destinations or state specific needs. Moreover, local navigation trajectories overlook global context and high-level task planning. To address these issues, we propose NavRAG, a retrieval-augmented generation (RAG) framework that generates user demand instructions for VLN. NavRAG leverages LLM to build a hierarchical scene description tree for 3D scene understanding from global layout to local details, then simulates various user roles with specific demands to retrieve from the scene tree, generating diverse instructions with LLM. We annotate over 2 million navigation instructions across 861 scenes and evaluate the data quality and navigation performance of trained models.
♻ ☆ Speculative Decoding and Beyond: An In-Depth Survey of Techniques
Sequential dependencies present a fundamental bottleneck in deploying large-scale autoregressive models, particularly for real-time applications. While traditional optimization approaches like pruning and quantization often compromise model quality, recent advances in generation-refinement frameworks demonstrate that this trade-off can be significantly mitigated. This survey presents a comprehensive taxonomy of generation-refinement frameworks, analyzing methods across autoregressive sequence tasks. We categorize methods based on their generation strategies (from simple n-gram prediction to sophisticated draft models) and refinement mechanisms (including single-pass verification and iterative approaches). Through systematic analysis of both algorithmic innovations and system-level implementations, we examine deployment strategies across computing environments and explore applications spanning text, images, and speech generation. This systematic examination of both theoretical frameworks and practical implementations provides a foundation for future research in efficient autoregressive decoding.
♻ ☆ Variational Best-of-N Alignment
Best-of-N (BoN) is a popular and effective algorithm for aligning language models to human preferences. The algorithm works as follows: at inference time, N samples are drawn from the language model, and the sample with the highest reward, as judged by a reward model, is returned as the output. Despite its effectiveness, BoN is computationally expensive; it reduces sampling throughput by a factor of N. To make BoN more efficient at inference time, one strategy is to fine-tune the language model to mimic what BoN does during inference. To achieve this, we derive the distribution induced by the BoN algorithm. We then propose to fine-tune the language model to minimize backward KL divergence to the BoN distribution. Our approach is analogous to mean-field variational inference and, thus, we term it variational BoN (vBoN). To the extent this fine-tuning is successful and we end up with a good approximation, we have reduced the inference cost by a factor of N. Our experiments on controlled generation and summarization tasks show that BoN is the most effective alignment method, and our variational approximation to BoN achieves the closest performance to BoN and surpasses models fine-tuned using the standard KL-constrained RL objective. In the controlled generation task, vBoN appears more frequently on the Pareto frontier of reward and KL divergence compared to other alignment methods. In the summarization task, vBoN achieves high reward values across various sampling temperatures.
♻ ☆ Exploring Iterative Controllable Summarization with Large Language Models
Large language models (LLMs) have demonstrated remarkable performance in abstractive summarization tasks. However, their ability to precisely control summary attributes (e.g., length or topic) remains underexplored, limiting their adaptability to specific user preferences. In this paper, we systematically explore the controllability of LLMs. To this end, we revisit summary attribute measurements and introduce iterative evaluation metrics, failure rate and average iteration count to precisely evaluate controllability of LLMs, rather than merely assessing errors. Our findings show that LLMs struggle more with numerical attributes than with linguistic attributes. To address this challenge, we propose a guide-to-explain framework (GTE) for controllable summarization. Our GTE framework enables the model to identify misaligned attributes in the initial draft and guides it in self-explaining errors in the previous output. By allowing the model to reflect on its misalignment, GTE generates well-adjusted summaries that satisfy the desired attributes with robust effectiveness, requiring surprisingly fewer iterations than other iterative approaches.
♻ ☆ "Nuclear Deployed!": Analyzing Catastrophic Risks in Decision-making of Autonomous LLM Agents
Large language models (LLMs) are evolving into autonomous decision-makers, raising concerns about catastrophic risks in high-stakes scenarios, particularly in Chemical, Biological, Radiological and Nuclear (CBRN) domains. Based on the insight that such risks can originate from trade-offs between the agent's Helpful, Harmlessness and Honest (HHH) goals, we build a novel three-stage evaluation framework, which is carefully constructed to effectively and naturally expose such risks. We conduct 14,400 agentic simulations across 12 advanced LLMs, with extensive experiments and analysis. Results reveal that LLM agents can autonomously engage in catastrophic behaviors and deception, without being deliberately induced. Furthermore, stronger reasoning abilities often increase, rather than mitigate, these risks. We also show that these agents can violate instructions and superior commands. On the whole, we empirically prove the existence of catastrophic risks in autonomous LLM agents. We will release our code upon request.
comment: Please visit https://llm-catastrophic-risks.github.io for a quick tour of our project
♻ ☆ LaERC-S: Improving LLM-based Emotion Recognition in Conversation with Speaker Characteristics COLING 2025
Emotion recognition in conversation (ERC), the task of discerning human emotions for each utterance within a conversation, has garnered significant attention in human-computer interaction systems. Previous ERC studies focus on speaker-specific information that predominantly stems from relationships among utterances, which lacks sufficient information around conversations. Recent research in ERC has sought to exploit pre-trained large language models (LLMs) with speaker modelling to comprehend emotional states. Although these methods have achieved encouraging results, the extracted speaker-specific information struggles to indicate emotional dynamics. In this paper, motivated by the fact that speaker characteristics play a crucial role and LLMs have rich world knowledge, we present LaERC-S, a novel framework that stimulates LLMs to explore speaker characteristics involving the mental state and behavior of interlocutors, for accurate emotion predictions. To endow LLMs with this knowledge information, we adopt the two-stage learning to make the models reason speaker characteristics and track the emotion of the speaker in complex conversation scenarios. Extensive experiments on three benchmark datasets demonstrate the superiority of LaERC-S, reaching the new state-of-the-art.
comment: COLING 2025
♻ ☆ Enhancing Large Language Models with Pseudo- and Multisource- Knowledge Graphs for Open-ended Question Answering
Mitigating the hallucinations of Large Language Models is a crucial task. Although some existing methods employ self-enhancement techniques, they fall short of effectively addressing unknown factual hallucinations. Meanwhile, Knowledge Graph (KG) enhancement approaches fail to address the generalization across different KG sources and the enhancement of open-ended answer questions simultaneously. To tackle these limitations, we propose a framework that combines Pseudo-Graph Generation and Atomic Knowledge Verification (PG\&AKV). Enhancement of open-ended question-answering begins with leveraging the Pseudo-Graph Generation to provide the related knowledge framework. Subsequently, Atomic Knowledge Verification utilizes atomic-level knowledge querying and verification to achieve generalizability under different KG sources. Compared to the baseline, this approach yields a minimum improvement of 11.5 in the ROUGE-L score for open-ended questions. For precise-answered questions, we observe a minimum accuracy improvement of 7.5%. Moreover, PG\&AKV also exhibits generalizability across different KG sources. Utilizing KG different from the question sources, PG\&AKV can even achieve at least a 3.5 % performance improvement. In summary, our results pave the way for enhancing LLMs by incorporating Pseudo- and Multisource-KGs, particularly in the filed of open-ended questions.
♻ ☆ ECLeKTic: a Novel Challenge Set for Evaluation of Cross-Lingual Knowledge Transfer
To achieve equitable performance across languages, multilingual large language models (LLMs) must be able to abstract knowledge beyond the language in which it was acquired. However, the current literature lacks reliable ways to measure LLMs' capability of cross-lingual knowledge transfer. To that end, we present ECLeKTic, a multilingual closed-book QA (CBQA) dataset that Evaluates Cross-Lingual Knowledge Transfer in a simple, black-box manner. We detected information with uneven coverage across languages by controlling for presence and absence of Wikipedia articles in 12 languages. We generated knowledge-seeking questions in a source language, for which the answer appears in a relevant Wikipedia article and translated them to all other 11 languages, for which the respective Wikipedias lack equivalent articles. Assuming that Wikipedia reflects the prominent knowledge in the LLM's training data, to solve ECLeKTic's CBQA task the model is required to transfer knowledge between languages. Experimenting with 8 LLMs, we show that SOTA models struggle to effectively share knowledge across, languages even if they can predict the answer well for queries in the same language the knowledge was acquired in.
♻ ☆ SURGE: On the Potential of Large Language Models as General-Purpose Surrogate Code Executors
Neural surrogate models have emerged as powerful and efficient tools in data mining. Meanwhile, large language models (LLMs) have demonstrated remarkable capabilities in code-related tasks. We investigate a novel application: using LLMs as surrogate models for code execution prediction. Given LLMs' unique ability to understand and process diverse programs, they present a promising direction for building general-purpose surrogate models. To systematically investigate this capability, we introduce SURGE, a comprehensive benchmark with $1160$ problems covering $8$ key aspects: multi-language programming tasks, competition-level programming problems, repository-level code analysis, high-cost scientific computing, time-complexity-intensive algorithms, buggy code analysis, programs dependent on specific compilers or execution environments, and formal mathematical proof verification. Through extensive empirical analysis of $21$ open-source and proprietary LLMs, we examine scaling laws, data efficiency, and predictive accuracy. Our findings reveal important insights about the feasibility of LLMs as efficient surrogates for computational processes, with implications for automated software testing, program analysis, and computational resource optimization in data mining applications. Code and dataset are released at https://github.com/Imbernoulli/SURGE.
♻ ☆ Robust Preference Optimization through Reward Model Distillation
Language model (LM) post-training (or alignment) involves maximizing a reward function that is derived from preference annotations. Direct Preference Optimization (DPO) is a popular offline alignment method that trains a policy directly on preference data without the need to train a reward model or apply reinforcement learning. However, the empirical evidence suggests that DPO typically assigns implicit rewards that overfit, and trend towards infinite magnitude. This frequently leads to degenerate policies, sometimes causing even the probabilities of the preferred generations to go to zero. In this work, we analyze this phenomenon and use distillation to get a better proxy for the true preference distribution over generation pairs: we train the LM such that its induced implicit reward, i.e., the scaled log-likelihood ratio of the model to the reference model, matches an explicit reward model trained on the preference data. Moreover, to account for uncertainty in the reward model we are distilling from, we optimize against a family of reward models that, as a whole, is likely to include at least one reasonable proxy for the preference distribution. Our results show that distilling from such a family of reward models leads to improved robustness to distribution shift in preference annotations, while preserving the simple supervised nature of DPO.
♻ ☆ Dynamics of Instruction Fine-Tuning for Chinese Large Language Models COLING 2025
Instruction tuning is a burgeoning method to elicit the general intelligence of Large Language Models (LLMs). While numerous studies have examined the impact of factors such as data volume and model size on English models, the scaling properties of instruction tuning in other languages remain largely unexplored. In this work, we systematically investigate the effects of data quantity, model size, and data construction methods on instruction tuning for Chinese LLMs. We utilize a newly curated dataset, DoIT, which includes over 40,000 high-quality instruction instances covering ten underlying abilities, such as creative writing, code generation, and logical reasoning. Our experiments, conducted on models ranging from 7b to 33b parameters, yield three key findings: (i) While these factors directly affect overall model performance, some abilities are more responsive to scaling, whereas others demonstrate significant resistance. (ii) The scaling sensitivity of different abilities to these factors can be explained by two features: Complexity and Transference. (iii) By tailoring training strategies to their varying sensitivities, specific abilities can be efficiently learned, enhancing performance on two public benchmarks.
comment: Accepted to COLING 2025
♻ ☆ Selected Languages are All You Need for Cross-lingual Truthfulness Transfer COLING2025
Truthfulness stands out as an essential challenge for Large Language Models (LLMs). Although many works have developed various ways for truthfulness enhancement, they seldom focus on truthfulness in multilingual scenarios. Meanwhile, contemporary multilingual aligning technologies struggle to balance numerous languages and often exhibit serious truthfulness gaps across different languages, especially those that differ greatly from English. In our work, we extend truthfulness evaluation to multilingual contexts and propose a practical method for cross-lingual truthfulness transfer called Fact-aware Multilingual Selective Synergy (FaMSS). FaMSS is able to select an optimal subset of all tested languages by language bias and transfer contributions, and then employ translation instruction tuning for cross-lingual truthfulness transfer. Experimental results demonstrate that our approach can effectively reduce the multilingual representation disparity and boost cross-lingual truthfulness transfer of LLMs.
comment: 16 pages, COLING2025
♻ ☆ Enabling Auditory Large Language Models for Automatic Speech Quality Evaluation ICASSP 2025
Speech quality assessment typically requires evaluating audio from multiple aspects, such as mean opinion score (MOS) and speaker similarity (SIM) \etc., which can be challenging to cover using one small model designed for a single task. In this paper, we propose leveraging recently introduced auditory large language models (LLMs) for automatic speech quality assessment. By employing task-specific prompts, auditory LLMs are finetuned to predict MOS, SIM and A/B testing results, which are commonly used for evaluating text-to-speech systems. Additionally, the finetuned auditory LLM is able to generate natural language descriptions assessing aspects like noisiness, distortion, discontinuity, and overall quality, providing more interpretable outputs. Extensive experiments have been performed on the NISQA, BVCC, SOMOS and VoxSim speech quality datasets, using open-source auditory LLMs such as SALMONN, Qwen-Audio, and Qwen2-Audio. For the natural language descriptions task, a commercial model Google Gemini 1.5 Pro is also evaluated. The results demonstrate that auditory LLMs achieve competitive performance compared to state-of-the-art task-specific small models in predicting MOS and SIM, while also delivering promising results in A/B testing and natural language descriptions. Our data processing scripts and finetuned model checkpoints can be found at https://github.com/bytedance/SALMONN.
comment: Accepted by ICASSP 2025
♻ ☆ DailyDilemmas: Revealing Value Preferences of LLMs with Quandaries of Daily Life ICLR 2025
As users increasingly seek guidance from LLMs for decision-making in daily life, many of these decisions are not clear-cut and depend significantly on the personal values and ethical standards of people. We present DailyDilemmas, a dataset of 1,360 moral dilemmas encountered in everyday life. Each dilemma presents two possible actions, along with affected parties and relevant human values for each action. Based on these dilemmas, we gather a repository of human values covering diverse everyday topics, such as interpersonal relationships, workplace, and environmental issues. With DailyDilemmas, we evaluate LLMs on these dilemmas to determine what action they will choose and the values represented by these action choices. Then, we analyze values through the lens of five theoretical frameworks inspired by sociology, psychology, and philosophy, including the World Values Survey, Moral Foundations Theory, Maslow's Hierarchy of Needs, Aristotle's Virtues, and Plutchik's Wheel of Emotions. For instance, we find LLMs are most aligned with self-expression over survival in World Values Survey and care over loyalty in Moral Foundations Theory. Interestingly, we find substantial preference differences in models for some core values. For example, for truthfulness, Mixtral-8x7B neglects it by 9.7% while GPT-4-turbo selects it by 9.4%. We also study the recent guidance released by OpenAI (ModelSpec), and Anthropic (Constitutional AI) to understand how their designated principles reflect their models' actual value prioritization when facing nuanced moral reasoning in daily-life settings. Finally, we find that end users cannot effectively steer such prioritization using system prompts.
comment: Accepted into ICLR 2025 (spotlight)
♻ ☆ Test-Time Compute: from System-1 Thinking to System-2 Thinking
The remarkable performance of the o1 model in complex reasoning demonstrates that test-time compute scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time compute scaling. We trace the concept of test-time compute back to System-1 models. In System-1 models, test-time compute addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time compute in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out a few possible future directions.
comment: work in progress
♻ ☆ Learning to Align Multi-Faceted Evaluation: A Unified and Robust Framework
Large Language Models (LLMs) are being used more and more extensively for automated evaluation in various scenarios. Previous studies have attempted to fine-tune open-source LLMs to replicate the evaluation explanations and judgments of powerful proprietary models, such as GPT-4. However, these methods are largely limited to text-based analyses under predefined general criteria, resulting in reduced adaptability for unseen instructions and demonstrating instability in evaluating adherence to quantitative and structural constraints. To address these limitations, we propose a novel evaluation framework, ARJudge, that adaptively formulates evaluation criteria and synthesizes both text-based and code-driven analyses to evaluate LLM responses. ARJudge consists of two components: a fine-tuned Analyzer that generates multi-faceted evaluation analyses and a tuning-free Refiner that combines and refines all analyses to make the final judgment. We construct a Composite Analysis Corpus that integrates tasks for evaluation criteria generation alongside text-based and code-driven analysis generation to train the Analyzer. Our results demonstrate that ARJudge outperforms existing fine-tuned evaluators in effectiveness and robustness. Furthermore, it demonstrates the importance of multi-faceted evaluation and code-driven analyses in enhancing evaluation capabilities.
♻ ☆ Subtle Errors Matter: Preference Learning via Error-injected Self-editing
Large Language Models (LLMs) have exhibited strong mathematical reasoning prowess, tackling tasks ranging from basic arithmetic to advanced competition-level problems. However, frequently occurring subtle yet critical errors, such as miscalculations or incorrect substitutions, limit the LLMs' full potential. Existing studies to improve mathematical ability typically involve applying preference learning to step-wise solution pairs. Although these methods leverage samples of varying granularity to mitigate reasoning errors, they overlook critical subtle errors. In this work, we propose a novel preference learning framework called eRror-Injected Self-Editing (RISE), which injects predefined subtle errors into pivotal tokens in reasoning or computation steps to construct hard pairs for error mitigation. In detail, RISE uses the LLM itself to edit a small number of tokens in the solution, injecting designed subtle errors. Then, pairs composed of self-edited solutions and their corresponding correct ones, along with pairs of correct and incorrect solutions obtained through sampling, are used together for subtle error-aware DPO training. Compared with other preference learning methods, RISE further refines the training objective without requiring fine-grained sampling or preference annotation. Extensive experiments validate the effectiveness of RISE, with preference learning on Qwen2-7B-Instruct yielding notable improvements of 3.0% on GSM8K and 7.9% on MATH with only 4.5K training samples. Moreover, the effect of error mitigation extends from mathematical reasoning to logical reasoning and code generation.
♻ ☆ Learning Efficient Recursive Numeral Systems via Reinforcement Learning
It has previously been shown that by using reinforcement learning (RL), agents can derive simple approximate and exact-restricted numeral systems that are similar to human ones (Carlsson, 2021). However, it is a major challenge to show how more complex recursive numeral systems, similar to for example English, could arise via a simple learning mechanism such as RL. Here, we introduce an approach towards deriving a mechanistic explanation of the emergence of efficient recursive number systems. We consider pairs of agents learning how to communicate about numerical quantities through a meta-grammar that can be gradually modified throughout the interactions. %We find that the seminal meta-grammar of Hurford (Hurford, 1975) is not suitable for this application as its optimization results in systems that deviate from standard conventions observed within human numeral systems. We propose a simple modification which addresses this issue. Utilising a slightly modified version of the meta-grammar of Hurford, we demonstrate that our RL agents, shaped by the pressures for efficient communication, can effectively modify their lexicon towards Pareto-optimal configurations which are comparable to those observed within human numeral systems in terms of their efficiency.
comment: 8 pages, 5 figures
♻ ☆ SpikeLLM: Scaling up Spiking Neural Network to Large Language Models via Saliency-based Spiking
Recent advancements in large language models (LLMs) with billions of parameters have improved performance in various applications, but their inference processes demand significant energy and computational resources. In contrast, the human brain, with approximately 86 billion neurons, is much more energy-efficient than LLMs with similar parameters. Inspired by this, we redesign 7$\sim$70 billion parameter LLMs using bio-plausible spiking mechanisms, emulating the efficient behavior of the human brain. We propose the first spiking large language model, SpikeLLM. Coupled with the proposed model, two essential approaches are proposed to improve spike training efficiency: Generalized Integrate-and-Fire (GIF) neurons to compress spike length from $T$ to $\frac{T}{L} \log_2 L$ bits, and an Optimal Brain Spiking framework to divide outlier channels and allocate different $T$ for GIF neurons, which further compresses spike length to approximate $log_2T$ bits. The necessity of spike-driven LLM is proved by comparison with quantized LLMs with similar operations. In the OmniQuant pipeline, SpikeLLM reduces 11.01% WikiText2 perplexity and improves 2.55% accuracy of common scene reasoning on a LLAMA-7B W4A4 model. In the GPTQ pipeline, SpikeLLM achieves direct additive in linear layers, significantly exceeding PB-LLMs.
♻ ☆ SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer NAACL 2025
Although pre-training achieves remarkable performance, it suffers from task-agnostic backdoor attacks due to vulnerabilities in data and training mechanisms. These attacks can transfer backdoors to various downstream tasks. In this paper, we introduce $\mathtt{maxEntropy}$, an entropy-based poisoning filter that mitigates such risks. To overcome the limitations of manual target setting and explicit triggers, we propose $\mathtt{SynGhost}$, an invisible and universal task-agnostic backdoor attack via syntactic transfer, further exposing vulnerabilities in pre-trained language models (PLMs). Specifically, $\mathtt{SynGhost}$ injects multiple syntactic backdoors into the pre-training space through corpus poisoning, while preserving the PLM's pre-training capabilities. Second, $\mathtt{SynGhost}$ adaptively selects optimal targets based on contrastive learning, creating a uniform distribution in the pre-training space. To identify syntactic differences, we also introduce an awareness module to minimize interference between backdoors. Experiments show that $\mathtt{SynGhost}$ poses significant threats and can transfer to various downstream tasks. Furthermore, $\mathtt{SynGhost}$ resists defenses based on perplexity, fine-pruning, and $\mathtt{maxEntropy}$. The code is available at https://github.com/Zhou-CyberSecurity-AI/SynGhost.
comment: 17 pages, 16 figures, 12 tables, accepted at NAACL 2025 Findings
♻ ☆ The Rise and Down of Babel Tower: Investigating the Evolution Process of Multilingual Code Large Language Model ICLR 2025
Large language models (LLMs) have shown significant multilingual capabilities. However, the mechanisms underlying the development of these capabilities during pre-training are not well understood. In this paper, we use code LLMs as an experimental platform to explore the evolution of multilingual capabilities in LLMs during the pre-training process. Based on our observations, we propose the Babel Tower Hypothesis, which describes the entire process of LLMs acquiring new language capabilities. During the learning process, multiple languages initially share a single knowledge system dominated by the primary language and gradually develop language-specific knowledge systems. We then validate the above hypothesis by tracking the internal states of the LLMs through identifying working languages and language transferring neurons. Experimental results show that the internal state changes of the LLM are consistent with our Babel Tower Hypothesis. Building on these insights, we propose a novel method to construct an optimized pre-training corpus for multilingual code LLMs, which significantly outperforms LLMs trained on the original corpus. The proposed Babel Tower Hypothesis provides new insights into designing pre-training data distributions to achieve optimal multilingual capabilities in LLMs.
comment: Accepted to ICLR 2025
♻ ☆ Order Matters: Investigate the Position Bias in Multi-constraint Instruction Following
Real-world instructions with multiple constraints pose a significant challenge to existing large language models (LLMs). An observation is that the LLMs exhibit dramatic performance fluctuation when disturbing the order of the incorporated constraints. Yet, none of the existing works has systematically investigated this position bias problem in the field of multi-constraint instruction following. To bridge this gap, we design a probing task where we quantitatively measure the difficulty distribution of the constraints by a novel Difficulty Distribution Index (CDDI). Through the experimental results, we find that LLMs are more performant when presented with the constraints in a ``hard-to-easy'' order. This preference can be generalized to LLMs with different architecture or different sizes of parameters. Additionally, we conduct an explanation study, providing an intuitive insight into the correlation between the LLM's attention and constraint orders. Our code and dataset are publicly available at https://github.com/meowpass/PBIF.
♻ ☆ Representation Engineering: A Top-Down Approach to AI Transparency
In this paper, we identify and characterize the emerging area of representation engineering (RepE), an approach to enhancing the transparency of AI systems that draws on insights from cognitive neuroscience. RepE places population-level representations, rather than neurons or circuits, at the center of analysis, equipping us with novel methods for monitoring and manipulating high-level cognitive phenomena in deep neural networks (DNNs). We provide baselines and an initial analysis of RepE techniques, showing that they offer simple yet effective solutions for improving our understanding and control of large language models. We showcase how these methods can provide traction on a wide range of safety-relevant problems, including honesty, harmlessness, power-seeking, and more, demonstrating the promise of top-down transparency research. We hope that this work catalyzes further exploration of RepE and fosters advancements in the transparency and safety of AI systems.
comment: Code is available at https://github.com/andyzoujm/representation-engineering
♻ ☆ TokenSelect: Efficient Long-Context Inference and Length Extrapolation for LLMs via Dynamic Token-Level KV Cache Selection
The rapid advancement of Large Language Models (LLMs) has driven growing demand for processing extended context sequences in contemporary applications. However, this progress faces two major challenges: performance degradation due to sequence lengths out-of-distribution, and excessively long inference times caused by the quadratic computational complexity of attention. These issues hinder the application of LLMs in long-context scenarios. In this paper, we propose Dynamic Token-Level KV Cache Selection (TokenSelect), a training-free method for efficient and accurate long-context inference. TokenSelect builds upon the observation of non-contiguous attention sparsity, using Query-Key dot products to measure per-head KV Cache criticality at token-level. By per-head soft voting mechanism, TokenSelect selectively involves a few critical KV cache tokens in attention calculation without sacrificing accuracy. To further accelerate TokenSelect, we design the Selection Cache based on observations of consecutive Query similarity and implemented efficient dot product kernel, significantly reducing the overhead. A comprehensive evaluation of TokenSelect demonstrates up to 23.84x speedup in attention computation and up to 2.28x acceleration in end-to-end latency, while providing superior performance compared to state-of-the-art long-context inference methods.
♻ ☆ Enhancing Conversational Agents with Theory of Mind: Aligning Beliefs, Desires, and Intentions for Human-Like Interaction
Natural language interaction with agentic Artificial Intelligence (AI), driven by Large Language Models (LLMs), is expected to remain a dominant paradigm in the near future. While humans instinctively align their communication with mental states -- an ability known as Theory of Mind (ToM), current LLM powered systems exhibit significant limitations in this regard. This study examines the extent to which open source language models (LLaMA) can capture and preserve ToM related information and how effectively it contributes to consistent ToM reasoning in generated responses. We further investigate whether explicit manipulation of ToM related components, such as beliefs, desires, and intentions, can enhance response alignment. Experiments on two LLaMA 3 variants demonstrate that incorporating ToM informed alignment improves response quality, achieving win rates of 67 and 63 percent for the 3B and 8B models, respectively. These findings highlight the potential of ToM driven strategies to improve alignment in LLM based conversational agents.
♻ ☆ Structural-Entropy-Based Sample Selection for Efficient and Effective Learning ICLR 2025
Sample selection improves the efficiency and effectiveness of machine learning models by providing informative and representative samples. Typically, samples can be modeled as a sample graph, where nodes are samples and edges represent their similarities. Most existing methods are based on local information, such as the training difficulty of samples, thereby overlooking global information, such as connectivity patterns. This oversight can result in suboptimal selection because global information is crucial for ensuring that the selected samples well represent the structural properties of the graph. To address this issue, we employ structural entropy to quantify global information and losslessly decompose it from the whole graph to individual nodes using the Shapley value. Based on the decomposition, we present $\textbf{S}$tructural-$\textbf{E}$ntropy-based sample $\textbf{S}$election ($\textbf{SES}$), a method that integrates both global and local information to select informative and representative samples. SES begins by constructing a $k$NN-graph among samples based on their similarities. It then measures sample importance by combining structural entropy (global metric) with training difficulty (local metric). Finally, SES applies importance-biased blue noise sampling to select a set of diverse and representative samples. Comprehensive experiments on three learning scenarios -- supervised learning, active learning, and continual learning -- clearly demonstrate the effectiveness of our method.
comment: Published as a conference paper at ICLR 2025
♻ ☆ Spontaneous Giving and Calculated Greed in Language Models
Large language models demonstrate advanced problem-solving capabilities by incorporating reasoning techniques such as chain of thought and reflection. However, how these reasoning capabilities extend to social intelligence remains unclear. In this study, we investigate this question using economic games that model social dilemmas, where social intelligence plays a crucial role. First, we examine the effects of chain-of-thought and reflection techniques in a public goods game. We then extend our analysis to six economic games on cooperation and punishment, comparing off-the-shelf non-reasoning and reasoning models. We find that reasoning models significantly reduce cooperation and norm enforcement, prioritizing individual rationality. Consequently, groups with more reasoning models exhibit less cooperation and lower gains through repeated interactions. These behaviors parallel human tendencies of "spontaneous giving and calculated greed." Our results suggest the need for AI architectures that incorporate social intelligence alongside reasoning capabilities to ensure that AI supports, rather than disrupts, human cooperative intuition.
♻ ☆ Generative Representational Instruction Tuning
All text-based language problems can be reduced to either generation or embedding. Current models only perform well at one or the other. We introduce generative representational instruction tuning (GRIT) whereby a large language model is trained to handle both generative and embedding tasks by distinguishing between them through instructions. Compared to other open models, our resulting GritLM 7B sets a new state of the art on the Massive Text Embedding Benchmark (MTEB) and outperforms all models up to its size on a range of generative tasks. By scaling up further, GritLM 8x7B outperforms all open generative language models that we tried while still being among the best embedding models. Notably, we find that GRIT matches training on only generative or embedding data, thus we can unify both at no performance loss. Among other benefits, the unification via GRIT speeds up Retrieval-Augmented Generation (RAG) by > 60% for long documents, by no longer requiring separate retrieval and generation models. Models, code, etc. are freely available at https://github.com/ContextualAI/gritlm.
comment: 67 pages (16 main), 25 figures, 34 tables
♻ ☆ Perturbation-Restrained Sequential Model Editing ICLR 2025
Model editing is an emerging field that focuses on updating the knowledge embedded within large language models (LLMs) without extensive retraining. However, current model editing methods significantly compromise the general abilities of LLMs as the number of edits increases, and this trade-off poses a substantial challenge to the continual learning of LLMs. In this paper, we first theoretically analyze that the factor affecting the general abilities in sequential model editing lies in the condition number of the edited matrix. The condition number of a matrix represents its numerical sensitivity, and therefore can be used to indicate the extent to which the original knowledge associations stored in LLMs are perturbed after editing. Subsequently, statistical findings demonstrate that the value of this factor becomes larger as the number of edits increases, thereby exacerbating the deterioration of general abilities. To this end, a framework termed Perturbation Restraint on Upper bouNd for Editing (PRUNE) is proposed, which applies the condition number restraints in sequential editing. These restraints can lower the upper bound on perturbation to edited models, thus preserving the general abilities. Systematically, we conduct experiments employing three editing methods on three LLMs across four downstream tasks. The results show that PRUNE can preserve general abilities while maintaining the editing performance effectively in sequential model editing. The code are available at https://github.com/mjy1111/PRUNE.
comment: Accepted by ICLR 2025
♻ ☆ A-MEM: Agentic Memory for LLM Agents
While large language model (LLM) agents can effectively use external tools for complex real-world tasks, they require memory systems to leverage historical experiences. Current memory systems enable basic storage and retrieval but lack sophisticated memory organization, despite recent attempts to incorporate graph databases. Moreover, these systems' fixed operations and structures limit their adaptability across diverse tasks. To address this limitation, this paper proposes a novel agentic memory system for LLM agents that can dynamically organize memories in an agentic way. Following the basic principles of the Zettelkasten method, we designed our memory system to create interconnected knowledge networks through dynamic indexing and linking. When a new memory is added, we generate a comprehensive note containing multiple structured attributes, including contextual descriptions, keywords, and tags. The system then analyzes historical memories to identify relevant connections, establishing links where meaningful similarities exist. Additionally, this process enables memory evolution - as new memories are integrated, they can trigger updates to the contextual representations and attributes of existing historical memories, allowing the memory network to continuously refine its understanding. Our approach combines the structured organization principles of Zettelkasten with the flexibility of agent-driven decision making, allowing for more adaptive and context-aware memory management. Empirical experiments on six foundation models show superior improvement against existing SOTA baselines. The source code is available at https://github.com/WujiangXu/AgenticMemory.
♻ ☆ QUAD-LLM-MLTC: Large Language Models Ensemble Learning for Healthcare Text Multi-Label Classification
The escalating volume of collected healthcare textual data presents a unique challenge for automated Multi-Label Text Classification (MLTC), which is primarily due to the scarcity of annotated texts for training and their nuanced nature. Traditional machine learning models often fail to fully capture the array of expressed topics. However, Large Language Models (LLMs) have demonstrated remarkable effectiveness across numerous Natural Language Processing (NLP) tasks in various domains, which show impressive computational efficiency and suitability for unsupervised learning through prompt engineering. Consequently, these LLMs promise an effective MLTC of medical narratives. However, when dealing with various labels, different prompts can be relevant depending on the topic. To address these challenges, the proposed approach, QUAD-LLM-MLTC, leverages the strengths of four LLMs: GPT-4o, BERT, PEGASUS, and BART. QUAD-LLM-MLTC operates in a sequential pipeline in which BERT extracts key tokens, PEGASUS augments textual data, GPT-4o classifies, and BART provides topics' assignment probabilities, which results in four classifications, all in a 0-shot setting. The outputs are then combined using ensemble learning and processed through a meta-classifier to produce the final MLTC result. The approach is evaluated using three samples of annotated texts, which contrast it with traditional and single-model methods. The results show significant improvements across the majority of the topics in the classification's F1 score and consistency (F1 and Micro-F1 scores of 78.17% and 80.16% with standard deviations of 0.025 and 0.011, respectively). This research advances MLTC using LLMs and provides an efficient and scalable solution to rapidly categorize healthcare-related text data without further training.
♻ ☆ Iterative Nash Policy Optimization: Aligning LLMs with General Preferences via No-Regret Learning
Reinforcement Learning with Human Feedback (RLHF) has achieved great success in aligning large language models (LLMs) with human preferences. Prevalent RLHF approaches are reward-based, following the Bradley-Terry (BT) model assumption, which may not fully capture the complexity of human preferences. In this paper, we explore RLHF under a general preference framework and approach it from a game-theoretic perspective. Specifically, we formulate the problem as a two-player game and propose a novel online algorithm, iterative Nash policy optimization (INPO). The key idea is to let the policy play against itself via no-regret learning, thereby approximating the Nash policy. Unlike previous methods, INPO bypasses the need for estimating the expected win rate for individual responses, which typically incurs high computational or annotation costs. Instead, we introduce a new loss objective that is directly minimized over a preference dataset. We provide theoretical analysis for our approach and demonstrate its effectiveness through experiments on various representative benchmarks. With an LLaMA-3-8B-based SFT model, INPO achieves a 42.6% length-controlled win rate on AlpacaEval 2.0 and a 37.8% win rate on Arena-Hard, showing substantial improvement over the state-of-the-art online RLHF algorithms.
♻ ☆ A Survey on Vision-Language-Action Models for Embodied AI
Embodied AI is widely recognized as a key element of artificial general intelligence because it involves controlling embodied agents to perform tasks in the physical world. Building on the success of large language models and vision-language models, a new category of multimodal models -- referred to as vision-language-action models (VLAs) -- has emerged to address language-conditioned robotic tasks in embodied AI by leveraging their distinct ability to generate actions. In recent years, a myriad of VLAs have been developed, making it imperative to capture the rapidly evolving landscape through a comprehensive survey. To this end, we present the first survey on VLAs for embodied AI. This work provides a detailed taxonomy of VLAs, organized into three major lines of research. The first line focuses on individual components of VLAs. The second line is dedicated to developing control policies adept at predicting low-level actions. The third line comprises high-level task planners capable of decomposing long-horizon tasks into a sequence of subtasks, thereby guiding VLAs to follow more general user instructions. Furthermore, we provide an extensive summary of relevant resources, including datasets, simulators, and benchmarks. Finally, we discuss the challenges faced by VLAs and outline promising future directions in embodied AI.
comment: 16 pages, a survey of vision-language-action models
♻ ☆ On the Generalization and Adaptation Ability of Machine-Generated Text Detectors in Academic Writing
The rising popularity of large language models (LLMs) has raised concerns about machine-generated text (MGT), particularly in academic settings, where issues like plagiarism and misinformation are prevalent. As a result, developing a highly generalizable and adaptable MGT detection system has become an urgent priority. Given that LLMs are most commonly misused in academic writing, this work investigates the generalization and adaptation capabilities of MGT detectors in three key aspects specific to academic writing: First, we construct MGT-Acedemic, a large-scale dataset comprising over 336M tokens and 749K samples. MGT-Acedemic focuses on academic writing, featuring human-written texts (HWTs) and MGTs across STEM, Humanities, and Social Sciences, paired with an extensible code framework for efficient benchmarking. Second, we benchmark the performance of various detectors for binary classification and attribution tasks in both in-domain and cross-domain settings. This benchmark reveals the often-overlooked challenges of attribution tasks. Third, we introduce a novel attribution task where models have to adapt to new classes over time without (or with very limited) access to prior training data in both few-shot and many-shot scenarios. We implement eight different adapting techniques to improve the performance and highlight the inherent complexity of the task. Our findings provide insights into the generalization and adaptation ability of MGT detectors across diverse scenarios and lay the foundation for building robust, adaptive detection systems. The code framework is available at https://github.com/Y-L-LIU/MGTBench-2.0.
♻ ☆ MMed-RAG: Versatile Multimodal RAG System for Medical Vision Language Models ICLR 2025
Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection method, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in the factual accuracy of Med-LVLMs. Our data and code are available in https://github.com/richard-peng-xia/MMed-RAG.
comment: ICLR 2025
♻ ☆ HORAE: A Domain-Agnostic Modeling Language for Automating Multimodal Service Regulation
Artificial intelligence is rapidly encroaching on the field of service regulation. This work-in-progress article presents the design principles behind HORAE, a unified specification language to model multimodal regulation rules across a diverse set of domains. We show how HORAE facilitates an intelligent service regulation pipeline by further exploiting a fine-tuned large language model named HORAE that automates the HORAE modeling process, thereby yielding an end-to-end framework for fully automated intelligent service regulation.
♻ ☆ Controllable Context Sensitivity and the Knob Behind It ICLR 2025
When making predictions, a language model must trade off how much it relies on its context vs. its prior knowledge. Choosing how sensitive the model is to its context is a fundamental functionality, as it enables the model to excel at tasks like retrieval-augmented generation and question-answering. In this paper, we search for a knob which controls this sensitivity, determining whether language models answer from the context or their prior knowledge. To guide this search, we design a task for controllable context sensitivity. In this task, we first feed the model a context (Paris is in England) and a question (Where is Paris?); we then instruct the model to either use its prior or contextual knowledge and evaluate whether it generates the correct answer for both intents (either France or England). When fine-tuned on this task, instruction-tuned versions of Llama-3.1, Mistral-v0.3, and Gemma-2 can solve it with high accuracy (85-95%). Analyzing these high-performing models, we narrow down which layers may be important to context sensitivity using a novel linear time algorithm. Then, in each model, we identify a 1-D subspace in a single layer that encodes whether the model follows context or prior knowledge. Interestingly, while we identify this subspace in a fine-tuned model, we find that the exact same subspace serves as an effective knob in not only that model but also non-fine-tuned instruct and base models of that model family. Finally, we show a strong correlation between a model's performance and how distinctly it separates context-agreeing from context-ignoring answers in this subspace. These results suggest a single subspace facilitates how the model chooses between context and prior knowledge, hinting at a simple fundamental mechanism that controls this behavior.
comment: Published as a conference paper at ICLR 2025
♻ ☆ A Pilot Empirical Study on When and How to Use Knowledge Graphs as Retrieval Augmented Generation
The integration of Knowledge Graphs (KGs) into the Retrieval Augmented Generation (RAG) framework has attracted significant interest, with early studies showing promise in mitigating hallucinations and improving model accuracy. However, a systematic understanding and comparative analysis of the rapidly emerging KG-RAG methods are still lacking. This paper seeks to lay the foundation for systematically answering the question of when and how to use KG-RAG by analyzing their performance in various application scenarios associated with different technical configurations. After outlining the mind map using KG-RAG framework and summarizing its popular pipeline, we conduct a pilot empirical study of KG-RAG works to reimplement and evaluate 6 KG-RAG methods across 7 datasets in diverse scenarios, analyzing the impact of 9 KG-RAG configurations in combination with 17 LLMs. Our results underscore the critical role of appropriate application conditions and optimal configurations of KG-RAG components.
comment: 8 pages, 2 figures, 14 tables
♻ ☆ A Closer Look at Machine Unlearning for Large Language Models ICLR 2025
Large language models (LLMs) may memorize sensitive or copyrighted content, raising privacy and legal concerns. Due to the high cost of retraining from scratch, researchers attempt to employ machine unlearning to remove specific content from LLMs while preserving the overall performance. In this paper, we discuss several issues in machine unlearning for LLMs and provide our insights on possible approaches. To address the issue of inadequate evaluation of model outputs after unlearning, we introduce three additional metrics to evaluate token diversity, sentence semantics, and factual correctness. We then categorize unlearning methods into untargeted and targeted, and discuss their issues respectively. Specifically, the behavior that untargeted unlearning attempts to approximate is unpredictable and may involve hallucinations, and existing regularization is insufficient for targeted unlearning. To alleviate these issues, we propose using the objective of maximizing entropy (ME) for untargeted unlearning and incorporate answer preservation (AP) loss as regularization for targeted unlearning. Experimental results across three scenarios, i.e., fictitious unlearning, continual unlearning, and real-world unlearning, demonstrate the effectiveness of our approaches. The code is available at https://github.com/sail-sg/closer-look-LLM-unlearning.
comment: ICLR 2025
♻ ☆ Exploring Adversarial Robustness in Classification tasks using DNA Language Models
DNA Language Models, such as GROVER, DNABERT2 and the Nucleotide Transformer, operate on DNA sequences that inherently contain sequencing errors, mutations, and laboratory-induced noise, which may significantly impact model performance. Despite the importance of this issue, the robustness of DNA language models remains largely underexplored. In this paper, we comprehensivly investigate their robustness in DNA classification by applying various adversarial attack strategies: the character (nucleotide substitutions), word (codon modifications), and sentence levels (back-translation-based transformations) to systematically analyze model vulnerabilities. Our results demonstrate that DNA language models are highly susceptible to adversarial attacks, leading to significant performance degradation. Furthermore, we explore adversarial training method as a defense mechanism, which enhances both robustness and classification accuracy. This study highlights the limitations of DNA language models and underscores the necessity of robustness in bioinformatics.
♻ ☆ Facilitating Multi-turn Function Calling for LLMs via Compositional Instruction Tuning ICLR 2025
Large Language Models (LLMs) have exhibited significant potential in performing diverse tasks, including the ability to call functions or use external tools to enhance their performance. While current research on function calling by LLMs primarily focuses on single-turn interactions, this paper addresses the overlooked necessity for LLMs to engage in multi-turn function calling--critical for handling compositional, real-world queries that require planning with functions but not only use functions. To facilitate this, we introduce an approach, BUTTON, which generates synthetic compositional instruction tuning data via bottom-up instruction construction and top-down trajectory generation. In the bottom-up phase, we generate simple atomic tasks based on real-world scenarios and build compositional tasks using heuristic strategies based on atomic tasks. Corresponding function definitions are then synthesized for these compositional tasks. The top-down phase features a multi-agent environment where interactions among simulated humans, assistants, and tools are utilized to gather multi-turn function calling trajectories. This approach ensures task compositionality and allows for effective function and trajectory generation by examining atomic tasks within compositional tasks. We produce a dataset BUTTONInstruct comprising 8k data points and demonstrate its effectiveness through extensive experiments across various LLMs.
comment: Accepted to ICLR 2025
♻ ☆ AdEval: Alignment-based Dynamic Evaluation to Mitigate Data Contamination in Large Language Models
As Large Language Models (LLMs) are pretrained on massive-scale corpora, the issue of data contamination has become increasingly severe, leading to potential overestimation of model performance during evaluation. To address this, we propose AdEval (Alignment-based Dynamic Evaluation), a dynamic data evaluation method aimed at mitigating the impact of data contamination on evaluation reliability. Experimental results on multiple datasets demonstrate that AdEval effectively reduces the impact of data contamination on evaluation outcomes, enhancing both the fairness and reliability of the evaluation process.
comment: There are serious academic problems in this paper, such as data falsification and plagiarism in the method of the paper
♻ ☆ OLMoE: Open Mixture-of-Experts Language Models
We introduce OLMoE, a fully open, state-of-the-art language model leveraging sparse Mixture-of-Experts (MoE). OLMoE-1B-7B has 7 billion (B) parameters but uses only 1B per input token. We pretrain it on 5 trillion tokens and further adapt it to create OLMoE-1B-7B-Instruct. Our models outperform all available models with similar active parameters, even surpassing larger ones like Llama2-13B-Chat and DeepSeekMoE-16B. We present various experiments on MoE training, analyze routing in our model showing high specialization, and open-source all aspects of our work: model weights, training data, code, and logs.
comment: 63 pages (24 main), 36 figures, 17 tables
♻ ☆ The Labyrinth of Links: Navigating the Associative Maze of Multi-modal LLMs ICLR 2025
Multi-modal Large Language Models (MLLMs) have exhibited impressive capability. However, recently many deficiencies of MLLMs have been found compared to human intelligence, $\textit{e.g.}$, hallucination. To drive the MLLMs study, the community dedicated efforts to building larger benchmarks with complex tasks. In this paper, we propose benchmarking an essential but usually overlooked intelligence: $\textbf{association}$, a human's basic capability to link observation and prior practice memory. To comprehensively investigate MLLM's performance on the association, we formulate the association task and devise a standard benchmark based on adjective and verb semantic concepts. Instead of costly data annotation and curation, we propose a convenient $\textbf{annotation-free}$ construction method transforming the general dataset for our association tasks. Simultaneously, we devise a rigorous data refinement process to eliminate confusion in the raw dataset. Building on this database, we establish three levels of association tasks: single-step, synchronous, and asynchronous associations. Moreover, we conduct a comprehensive investigation into the MLLMs' zero-shot association capabilities, addressing multiple dimensions, including three distinct memory strategies, both open-source and closed-source MLLMs, cutting-edge Mixture-of-Experts (MoE) models, and the involvement of human experts. Our systematic investigation shows that current open-source MLLMs consistently exhibit poor capability in our association tasks, even the currently state-of-the-art GPT-4V(vision) also has a significant gap compared to humans. We believe our benchmark would pave the way for future MLLM studies. $\textit{Our data and code are available at:}$ https://mvig-rhos.com/llm_inception.
comment: Accepted by ICLR 2025. Project page: https://mvig-rhos.com/llm_inception
♻ ☆ NL2FOL: Translating Natural Language to First-Order Logic for Logical Fallacy Detection
Translating natural language into formal language such as First-Order Logic (FOL) is a foundational challenge in NLP with wide-ranging applications in automated reasoning, misinformation tracking, and knowledge validation. In this paper, we introduce Natural Language to First-Order Logic (NL2FOL), a framework to autoformalize natural language to FOL step by step using Large Language Models (LLMs). Our approach addresses key challenges in this translation process, including the integration of implicit background knowledge. By leveraging structured representations generated by NL2FOL, we use Satisfiability Modulo Theory (SMT) solvers to reason about the logical validity of natural language statements. We present logical fallacy detection as a case study to evaluate the efficacy of NL2FOL. Being neurosymbolic, our approach also provides interpretable insights into the reasoning process and demonstrates robustness without requiring model fine-tuning or labeled training data. Our framework achieves strong performance on multiple datasets. On the LOGIC dataset, NL2FOL achieves an F1-score of 78%, while generalizing effectively to the LOGICCLIMATE dataset with an F1-score of 80%.
♻ ☆ Forecasting Frontier Language Model Agent Capabilities
As Language Models (LMs) increasingly operate as autonomous agents, accurately forecasting their capabilities becomes crucial for societal preparedness. We evaluate six forecasting methods that predict downstream capabilities of LM agents. We use "one-step" approaches that predict benchmark scores from input metrics like compute or model release date directly or "two-step" approaches that first predict an intermediate metric like the principal component of cross-benchmark performance (PC-1) and human-evaluated competitive Elo ratings. We evaluate our forecasting methods by backtesting them on a dataset of 38 LMs from the OpenLLM 2 leaderboard. We then use the validated two-step approach (Release Date$\to$Elo$\to$Benchmark) to predict LM agent performance for frontier models on three benchmarks: SWE-Bench Verified (software development), Cybench (cybersecurity assessment), and RE-Bench (ML research engineering). Our forecast predicts that by the beginning of 2026, non-specialized LM agents with low capability elicitation will reach a success rate of 54% on SWE-Bench Verified, while state-of-the-art LM agents will reach an 87% success rate. Our approach does not account for recent advances in inference-compute scaling and might thus be too conservative.
♻ ☆ Chain of Draft: Thinking Faster by Writing Less
Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning. However, humans typically employ a more efficient strategy: drafting concise intermediate thoughts that capture only essential information. In this work, we propose Chain of Draft (CoD), a novel paradigm inspired by human cognitive processes, where LLMs generate minimalistic yet informative intermediate reasoning outputs while solving tasks. By reducing verbosity and focusing on critical insights, CoD matches or surpasses CoT in accuracy while using as little as only 7.6% of the tokens, significantly reducing cost and latency across various reasoning tasks. Our code and data are available at https://github.com/sileix/chain-of-draft.
♻ ☆ SensorQA: A Question Answering Benchmark for Daily-Life Monitoring
With the rapid growth in sensor data, effectively interpreting and interfacing with these data in a human-understandable way has become crucial. While existing research primarily focuses on learning classification models, fewer studies have explored how end users can actively extract useful insights from sensor data, often hindered by the lack of a proper dataset. To address this gap, we introduce SensorQA, the first human-created question-answering (QA) dataset for long-term time-series sensor data for daily life monitoring. SensorQA is created by human workers and includes 5.6K diverse and practical queries that reflect genuine human interests, paired with accurate answers derived from sensor data. We further establish benchmarks for state-of-the-art AI models on this dataset and evaluate their performance on typical edge devices. Our results reveal a gap between current models and optimal QA performance and efficiency, highlighting the need for new contributions. The dataset and code are available at: https://github.com/benjamin-reichman/SensorQA.
♻ ☆ On Memory Construction and Retrieval for Personalized Conversational Agents
To deliver coherent and personalized experiences in long-term conversations, existing approaches typically perform retrieval augmented response generation by constructing memory banks from conversation history at either the turn-level, session-level, or through summarization techniques.In this paper, we present two key findings: (1) The granularity of memory unit matters: turn-level, session-level, and summarization-based methods each exhibit limitations in both memory retrieval accuracy and the semantic quality of the retrieved content. (2) Prompt compression methods, such as LLMLingua-2, can effectively serve as a denoising mechanism, enhancing memory retrieval accuracy across different granularities. Building on these insights, we propose SeCom, a method that constructs the memory bank at segment level by introducing a conversation segmentation model that partitions long-term conversations into topically coherent segments, while applying compression based denoising on memory units to enhance memory retrieval. Experimental results show that SeCom exhibits a significant performance advantage over baselines on long-term conversation benchmarks LOCOMO and Long-MT-Bench+. Additionally, the proposed conversation segmentation method demonstrates superior performance on dialogue segmentation datasets such as DialSeg711, TIAGE, and SuperDialSeg.
comment: 10 pages, 5 figures, conference
♻ ☆ R1-T1: Fully Incentivizing Translation Capability in LLMs via Reasoning Learning
Despite recent breakthroughs in reasoning-enhanced large language models (LLMs) like DeepSeek-R1, incorporating inference-time reasoning into machine translation (MT), where human translators naturally employ structured, multi-layered reasoning chain-of-thoughts (CoTs), is yet underexplored. Existing methods either design a fixed CoT tailored for a specific MT sub-task (e.g., literature translation), or rely on synthesizing CoTs unaligned with humans, limiting their adaptability to diverse translation scenarios. This paper introduces R1-Translator (R1-T1), a novel framework to achieve inference-time reasoning for general MT via reinforcement learning (RL) with human-aligned CoTs comprising six common patterns. Our approach pioneers three innovations: (1) extending reasoning-based translation beyond MT sub-tasks to six languages and diverse tasks (e.g., legal/medical domain adaptation, idiom resolution); (2) formalizing six expert-curated CoT templates that mirror hybrid human strategies like context-aware paraphrasing and back translation; and (3) enabling self-evolving CoT discovery through RL. Experimental results indicate a steady translation performance improvement in 11 languages and 40 translation directions on Flores-101 test set, especially on the languages unseen from training.
♻ ☆ InductionBench: LLMs Fail in the Simplest Complexity Class
Large language models (LLMs) have shown remarkable improvements in reasoning and many existing benchmarks have been addressed by models such as o1 and o3 either fully or partially. However, a majority of these benchmarks emphasize deductive reasoning, including mathematical and coding tasks in which rules such as mathematical axioms or programming syntax are clearly defined, based on which LLMs can plan and apply these rules to arrive at a solution. In contrast, inductive reasoning, where one infers the underlying rules from observed data, remains less explored. Such inductive processes lie at the heart of scientific discovery, as they enable researchers to extract general principles from empirical observations. To assess whether LLMs possess this capacity, we introduce InductionBench, a new benchmark designed to evaluate the inductive reasoning ability of LLMs. Our experimental findings reveal that even the most advanced models available struggle to master the simplest complexity classes within the subregular hierarchy of functions, highlighting a notable deficiency in current LLMs' inductive reasoning capabilities. Coda and data are available https://github.com/Wenyueh/inductive_reasoning_benchmark.
comment: 24 pages, 7 figures
♻ ☆ Can Knowledge Editing Really Correct Hallucinations? ICLR 2025
Large Language Models (LLMs) suffer from hallucinations, referring to the non-factual information in generated content, despite their superior capacities across tasks. Meanwhile, knowledge editing has been developed as a new popular paradigm to correct erroneous factual knowledge encoded in LLMs with the advantage of avoiding retraining from scratch. However, a common issue of existing evaluation datasets for knowledge editing is that they do not ensure that LLMs actually generate hallucinated answers to the evaluation questions before editing. When LLMs are evaluated on such datasets after being edited by different techniques, it is hard to directly adopt the performance to assess the effectiveness of different knowledge editing methods in correcting hallucinations. Thus, the fundamental question remains insufficiently validated: Can knowledge editing really correct hallucinations in LLMs? We proposed HalluEditBench to holistically benchmark knowledge editing methods in correcting real-world hallucinations. First, we rigorously construct a massive hallucination dataset with 9 domains, 26 topics and more than 6,000 hallucinations. Then, we assess the performance of knowledge editing methods in a holistic way on five dimensions including Efficacy, Generalization, Portability, Locality, and Robustness. Through HalluEditBench, we have provided new insights into the potentials and limitations of different knowledge editing methods in correcting hallucinations, which could inspire future improvements and facilitate progress in the field of knowledge editing.
comment: ICLR 2025. Main paper: 10 pages; total: 34 pages (including appendix). The first two authors contributed equally to this work. Code, data, results, and additional resources are available on the project website: https://llm-editing.github.io
♻ ☆ First-Person Fairness in Chatbots ICLR 2025
Evaluating chatbot fairness is crucial given their rapid proliferation, yet typical chatbot tasks (e.g., resume writing, entertainment) diverge from the institutional decision-making tasks (e.g., resume screening) which have traditionally been central to discussion of algorithmic fairness. The open-ended nature and diverse use-cases of chatbots necessitate novel methods for bias assessment. This paper addresses these challenges by introducing a scalable counterfactual approach to evaluate "first-person fairness," meaning fairness toward chatbot users based on demographic characteristics. Our method employs a Language Model as a Research Assistant (LMRA) to yield quantitative measures of harmful stereotypes and qualitative analyses of demographic differences in chatbot responses. We apply this approach to assess biases in six of our language models across millions of interactions, covering sixty-six tasks in nine domains and spanning two genders and four races. Independent human annotations corroborate the LMRA-generated bias evaluations. This study represents the first large-scale fairness evaluation based on real-world chat data. We highlight that post-training reinforcement learning techniques significantly mitigate these biases. This evaluation provides a practical methodology for ongoing bias monitoring and mitigation.
comment: In ICLR 2025, 59 pages, 27 figures
♻ ☆ Gumbel Counterfactual Generation From Language Models ICLR 2025
Understanding and manipulating the causal generation mechanisms in language models is essential for controlling their behavior. Previous work has primarily relied on techniques such as representation surgery -- e.g., model ablations or manipulation of linear subspaces tied to specific concepts -- to \emph{intervene} on these models. To understand the impact of interventions precisely, it is useful to examine \emph{counterfactuals} -- e.g., how a given sentence would have appeared had it been generated by the model following a specific intervention. We highlight that counterfactual reasoning is conceptually distinct from interventions, as articulated in Pearl's causal hierarchy. Based on this observation, we propose a framework for generating true string counterfactuals by reformulating language models as a structural equation model using the Gumbel-max trick, which we called Gumbel counterfactual generation. This reformulation allows us to model the joint distribution over original strings and their counterfactuals resulting from the same instantiation of the sampling noise. We develop an algorithm based on hindsight Gumbel sampling that allows us to infer the latent noise variables and generate counterfactuals of observed strings. Our experiments demonstrate that the approach produces meaningful counterfactuals while at the same time showing that commonly used intervention techniques have considerable undesired side effects.
comment: Accepted in ICLR 2025
♻ ☆ From Tokens to Words: On the Inner Lexicon of LLMs ICLR
Natural language is composed of words, but modern large language models (LLMs) process sub-words as input. A natural question raised by this discrepancy is whether LLMs encode words internally, and if so how. We present evidence that LLMs engage in an intrinsic detokenization process, where sub-word sequences are combined into coherent whole-word representations at their last token. Our experiments show that this process primarily takes place within the early and middle layers of the model. We further demonstrate its robustness to arbitrary splits (e.g., "cats" to "ca" and "ts"), typos, and importantly-to out-of-vocabulary words: when feeding the last token internal representations of such words to the model as input, it can "understand" them as the complete word despite never seeing such representations as input during training. Our findings suggest that LLMs maintain a latent vocabulary beyond the tokenizer's scope. These insights provide a practical, finetuning-free application for expanding the vocabulary of pre-trained models. By enabling the addition of new vocabulary words, we reduce input length and inference iterations, which reduces both space and model latency, with little to no loss in model accuracy.
comment: Accepted to the International Conference on Learning Representations (ICLR) 2025
♻ ☆ Naturally Occurring Feedback is Common, Extractable and Useful
Human feedback data is a critical component in developing language models. However, collecting this feedback is costly and ultimately not scalable. Inspired by the way human interlocutors provide spontaneous unsolicited feedback to each other, we propose to extract feedback that users naturally include when interacting with chat models. We manually annotated conversations to confirm the presence of naturally occurring feedback in a standard corpus, finding that as much as 30% of the chats include explicit feedback. Comparing to older datasets, we find that naturally occurring feedback is more prevalent in recent conversation datasets, suggesting that more than ever, naturally occurring feedback can serve as a valuable resource for feedback data. We propose a method for automatically extracting this feedback, and apply it to over 1M conversations to obtain hundreds of thousands of feedback samples. The extracted feedback shows promise: training with it improves over baseline models and enhances model alignment to human preferences.
♻ ☆ Towards Better Chain-of-Thought: A Reflection on Effectiveness and Faithfulness
Chain-of-thought (CoT) prompting demonstrates varying performance under different reasoning tasks. Previous work attempts to evaluate it but falls short in providing an in-depth analysis of patterns that influence the CoT. In this paper, we study the CoT performance from the perspective of effectiveness and faithfulness. For the former, we identify key factors that influence CoT effectiveness on performance improvement, including problem difficulty, information gain, and information flow. For the latter, we interpret the unfaithful CoT issue by conducting a joint analysis of the information interaction among the question, CoT, and answer. The result demonstrates that, when the LLM predicts answers, it can recall correct information missing in the CoT from the question, leading to the problem. Finally, we propose a novel algorithm to mitigate this issue, in which we recall extra information from the question to enhance the CoT generation and evaluate CoTs based on their information gain. Extensive experiments demonstrate that our approach enhances both the faithfulness and effectiveness of CoT.
comment: 18 pages, under review
♻ ☆ Surgical, Cheap, and Flexible: Mitigating False Refusal in Language Models via Single Vector Ablation ICLR 2025
Training a language model to be both helpful and harmless requires careful calibration of refusal behaviours: Models should refuse to follow malicious instructions or give harmful advice (e.g."how do I kill someone?"), but they should not refuse safe requests, even if they superficially resemble unsafe ones (e.g. "how do I kill a Python process?"). Avoiding such false refusal, as prior work has shown, is challenging even for highly-capable language models. In this paper, we propose a simple and surgical method for mitigating false refusal in language models via single vector ablation. For a given model, we extract a false refusal vector and show that ablating this vector reduces false refusal rate while preserving the model's safety and general capabilities. We also show that our approach can be used for fine-grained calibration of model safety. Our approach is training-free and model-agnostic, making it useful for mitigating the problem of false refusal in current and future language models.
comment: ICLR 2025
♻ ☆ Converting Transformers into DGNNs Form
Recent advances in deep learning have established Transformer architectures as the predominant modeling paradigm. Central to the success of Transformers is the self-attention mechanism, which scores the similarity between query and key matrices to modulate a value matrix. This operation bears striking similarities to digraph convolution, prompting an investigation into whether digraph convolution could serve as an alternative to self-attention. In this study, we formalize this concept by introducing a synthetic unitary digraph convolution based on the digraph Fourier transform. The resulting model, which we term Converter, effectively converts a Transformer into a Directed Graph Neural Network (DGNN) form. We have tested Converter on Long-Range Arena benchmark, long document classification, and DNA sequence-based taxonomy classification. Our experimental results demonstrate that Converter achieves superior performance while maintaining computational efficiency and architectural simplicity, which establishes it as a lightweight yet powerful Transformer variant.
comment: 21 pages, 3 figures, and 8 tables; pseudocode improved
♻ ☆ RAGEval: Scenario Specific RAG Evaluation Dataset Generation Framework
Retrieval-Augmented Generation (RAG) is a powerful approach that enables large language models (LLMs) to incorporate external knowledge. However, evaluating the effectiveness of RAG systems in specialized scenarios remains challenging due to the high costs of data construction and the lack of suitable evaluation metrics. This paper introduces RAGEval, a framework designed to assess RAG systems across diverse scenarios by generating high-quality documents, questions, answers, and references through a schema-based pipeline. With a focus on factual accuracy, we propose three novel metrics: Completeness, Hallucination, and Irrelevance to evaluate LLM generated responses rigorously. Experimental results show that RAGEval outperforms zero-shot and one-shot methods in terms of clarity, safety, conformity, and richness of generated samples. Furthermore, the use of LLMs for scoring the proposed metrics demonstrates a high level of consistency with human evaluations. RAGEval establishes a new paradigm for evaluating RAG systems in real-world applications. The code and dataset are released at https://github.com/OpenBMB/RAGEval.
comment: https://github.com/OpenBMB/RAGEval
♻ ☆ Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements ICLR 2025
The current paradigm for safety alignment of large language models (LLMs) follows a one-size-fits-all approach: the model refuses to interact with any content deemed unsafe by the model provider. This approach lacks flexibility in the face of varying social norms across cultures and regions. In addition, users may have diverse safety needs, making a model with static safety standards too restrictive to be useful, as well as too costly to be re-aligned. We propose Controllable Safety Alignment (CoSA), a framework designed to adapt models to diverse safety requirements without re-training. Instead of aligning a fixed model, we align models to follow safety configs -- free-form natural language descriptions of the desired safety behaviors -- that are provided as part of the system prompt. To adjust model safety behavior, authorized users only need to modify such safety configs at inference time. To enable that, we propose CoSAlign, a data-centric method for aligning LLMs to easily adapt to diverse safety configs. Furthermore, we devise a novel controllability evaluation protocol that considers both helpfulness and configured safety, summarizing them into CoSA-Score, and construct CoSApien, a human-authored benchmark that consists of real-world LLM use cases with diverse safety requirements and corresponding evaluation prompts. We show that CoSAlign leads to substantial gains of controllability over strong baselines including in-context alignment. Our framework encourages better representation and adaptation to pluralistic human values in LLMs, and thereby increasing their practicality.
comment: ICLR 2025 camera ready
♻ ☆ GlotCC: An Open Broad-Coverage CommonCrawl Corpus and Pipeline for Minority Languages NeurIPS 2024
The need for large text corpora has increased with the advent of pretrained language models and, in particular, the discovery of scaling laws for these models. Most available corpora have sufficient data only for languages with large dominant communities. However, there is no corpus available that (i) covers a wide range of minority languages; (ii) is generated by an open-source reproducible pipeline; and (iii) is rigorously cleaned from noise, making it trustworthy to use. We present GlotCC, a clean, document-level, 2TB general domain corpus derived from CommonCrawl, covering more than 1000 languages. We make GlotCC and the system used to generate it - including the pipeline, language identification model, and filters - available to the research community. Corpus v. 1.0 https://huggingface.co/datasets/cis-lmu/GlotCC-v1, Pipeline v. 3.0 https://github.com/cisnlp/GlotCC.
comment: NeurIPS 2024
♻ ☆ Talking Heads: Understanding Inter-layer Communication in Transformer Language Models
Although it is known that transformer language models (LMs) pass features from early layers to later layers, it is not well understood how this information is represented and routed by the model. We analyze a mechanism used in two LMs to selectively inhibit items in a context in one task, and find that it underlies a commonly used abstraction across many context-retrieval behaviors. Specifically, we find that models write into low-rank subspaces of the residual stream to represent features which are then read out by later layers, forming low-rank communication channels (Elhage et al., 2021) between layers. A particular 3D subspace in model activations in GPT-2 can be traversed to positionally index items in lists, and we show that this mechanism can explain an otherwise arbitrary-seeming sensitivity of the model to the order of items in the prompt. That is, the model has trouble copying the correct information from context when many items ``crowd" this limited space. By decomposing attention heads with the Singular Value Decomposition (SVD), we find that previously described interactions between heads separated by one or more layers can be predicted via analysis of their weight matrices alone. We show that it is possible to manipulate the internal model representations as well as edit model weights based on the mechanism we discover in order to significantly improve performance on our synthetic Laundry List task, which requires recall from a list, often improving task accuracy by over 20%. Our analysis reveals a surprisingly intricate interpretable structure learned from language model pretraining, and helps us understand why sophisticated LMs sometimes fail in simple domains, facilitating future analysis of more complex behaviors.
comment: Neurips 2024
♻ ☆ AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders
Fine-grained steering of language model outputs is essential for safety and reliability. Prompting and finetuning are widely used to achieve these goals, but interpretability researchers have proposed a variety of representation-based techniques as well, including sparse autoencoders (SAEs), linear artificial tomography, supervised steering vectors, linear probes, and representation finetuning. At present, there is no benchmark for making direct comparisons between these proposals. Therefore, we introduce AxBench, a large-scale benchmark for steering and concept detection, and report experiments on Gemma-2-2B and 9B. For steering, we find that prompting outperforms all existing methods, followed by finetuning. For concept detection, representation-based methods such as difference-in-means, perform the best. On both evaluations, SAEs are not competitive. We introduce a novel weakly-supervised representational method (Rank-1 Representation Finetuning; ReFT-r1), which is competitive on both tasks while providing the interpretability advantages that prompting lacks. Along with AxBench, we train and publicly release SAE-scale feature dictionaries for ReFT-r1 and DiffMean.
♻ ☆ De-identification is not enough: a comparison between de-identified and synthetic clinical notes
For sharing privacy-sensitive data, de-identification is commonly regarded as adequate for safeguarding privacy. Synthetic data is also being considered as a privacy-preserving alternative. Recent successes with numerical and tabular data generative models and the breakthroughs in large generative language models raise the question of whether synthetically generated clinical notes could be a viable alternative to real notes for research purposes. In this work, we demonstrated that (i) de-identification of real clinical notes does not protect records against a membership inference attack, (ii) proposed a novel approach to generate synthetic clinical notes using the current state-of-the-art large language models, (iii) evaluated the performance of the synthetically generated notes in a clinical domain task, and (iv) proposed a way to mount a membership inference attack where the target model is trained with synthetic data. We observed that when synthetically generated notes closely match the performance of real data, they also exhibit similar privacy concerns to the real data. Whether other approaches to synthetically generated clinical notes could offer better trade-offs and become a better alternative to sensitive real notes warrants further investigation.
comment: https://www.nature.com/articles/s41598-024-81170-y
♻ ☆ CPT-Boosted Wav2vec2.0: Towards Noise Robust Speech Recognition for Classroom Environments
Creating Automatic Speech Recognition (ASR) systems that are robust and resilient to classroom conditions is paramount to the development of AI tools to aid teachers and students. In this work, we study the efficacy of continued pretraining (CPT) in adapting Wav2vec2.0 to the classroom domain. We show that CPT is a powerful tool in that regard and reduces the Word Error Rate (WER) of Wav2vec2.0-based models by upwards of 10%. More specifically, CPT improves the model's robustness to different noises, microphones and classroom conditions.
comment: arXiv admin note: substantial text overlap with arXiv:2405.13018
Computer Vision and Pattern Recognition 70
♻ ☆ Mitigating Hallucinations in Large Vision-Language Models via DPO: On-Policy Data Hold the Key CVPR 2025
Hallucination remains a major challenge for Large Vision-Language Models (LVLMs). Direct Preference Optimization (DPO) has gained increasing attention as a simple solution to hallucination issues. It directly learns from constructed preference pairs that reflect the severity of hallucinations in responses to the same prompt and image. Nonetheless, different data construction methods in existing works bring notable performance variations. We identify a crucial factor here: outcomes are largely contingent on whether the constructed data aligns on-policy w.r.t the initial (reference) policy of DPO. Theoretical analysis suggests that learning from off-policy data is impeded by the presence of KL-divergence between the updated policy and the reference policy. From the perspective of dataset distribution, we systematically summarize the inherent flaws in existing algorithms that employ DPO to address hallucination issues. To alleviate the problems, we propose On-Policy Alignment (OPA)-DPO framework, which uniquely leverages expert feedback to correct hallucinated responses and aligns both the original and expert-revised responses in an on-policy manner. Notably, with only 4.8k data, OPA-DPO achieves an additional reduction in the hallucination rate of LLaVA-1.5-7B: 13.26% on the AMBER benchmark and 5.39% on the Object-Hal benchmark, compared to the previous SOTA algorithm trained with 16k samples. Our implementation is available at https://github.com/zhyang2226/OPA-DPO.
comment: Accepted by CVPR 2025
♻ ☆ Stereo Hand-Object Reconstruction for Human-to-Robot Handover
Jointly estimating hand and object shape facilitates the grasping task in human-to-robot handovers. However, relying on hand-crafted prior knowledge about the geometric structure of the object fails when generalising to unseen objects, and depth sensors fail to detect transparent objects such as drinking glasses. In this work, we propose a stereo-based method for hand-object reconstruction that combines single-view reconstructions probabilistically to form a coherent stereo reconstruction. We learn 3D shape priors from a large synthetic hand-object dataset to ensure that our method is generalisable, and use RGB inputs to better capture transparent objects. We show that our method reduces the object Chamfer distance compared to existing RGB based hand-object reconstruction methods on single view and stereo settings. We process the reconstructed hand-object shape with a projection-based outlier removal step and use the output to guide a human-to-robot handover pipeline with wide-baseline stereo RGB cameras. Our hand-object reconstruction enables a robot to successfully receive a diverse range of household objects from the human.
comment: 8 pages, 9 figures, 1 table
♻ ☆ EchoONE: Segmenting Multiple echocardiography Planes in One Model CVPR 2025
In clinical practice of echocardiography examinations, multiple planes containing the heart structures of different view are usually required in screening, diagnosis and treatment of cardiac disease. AI models for echocardiography have to be tailored for each specific plane due to the dramatic structure differences, thus resulting in repetition development and extra complexity. Effective solution for such a multi-plane segmentation (MPS) problem is highly demanded for medical images, yet has not been well investigated. In this paper, we propose a novel solution, EchoONE, for this problem with a SAM-based segmentation architecture, a prior-composable mask learning (PC-Mask) module for semantic-aware dense prompt generation, and a learnable CNN-branch with a simple yet effective local feature fusion and adaption (LFFA) module for SAM adapting. We extensively evaluated our method on multiple internal and external echocardiography datasets, and achieved consistently state-of-the-art performance for multi-source datasets with different heart planes. This is the first time that the MPS problem is solved in one model for echocardiography data. The code will be available at https://github.com/a2502503/EchoONE.
comment: Accepted by CVPR 2025
♻ ☆ Evaluating Intelligence via Trial and Error
Intelligence is a crucial trait for species to find solutions within a limited number of trial-and-error attempts. Building on this idea, we introduce Survival Game as a framework to evaluate intelligence based on the number of failed attempts in a trial-and-error process. Fewer failures indicate higher intelligence. When the expectation and variance of failure counts are both finite, it signals the ability to consistently find solutions to new challenges, which we define as the Autonomous Level of intelligence. Using Survival Game, we comprehensively evaluate existing AI systems. Our results show that while AI systems achieve the Autonomous Level in simple tasks, they are still far from it in more complex tasks, such as vision, search, recommendation, and language. While scaling current AI technologies might help, this would come at an astronomical cost. Projections suggest that achieving the Autonomous Level for general tasks would require $10^{26}$ parameters. To put this into perspective, loading such a massive model requires so many H100 GPUs that their total value is $10^{7}$ times that of Apple Inc.'s market value. Even with Moore's Law, supporting such a parameter scale would take $70$ years. This staggering cost highlights the complexity of human tasks and the inadequacies of current AI technologies. To further investigate this phenomenon, we conduct a theoretical analysis of Survival Game and its experimental results. Our findings suggest that human tasks possess a criticality property. As a result, Autonomous Level requires a deep understanding of the task's underlying mechanisms. Current AI systems, however, do not fully grasp these mechanisms and instead rely on superficial mimicry, making it difficult for them to reach an autonomous level. We believe Survival Game can not only guide the future development of AI but also offer profound insights into human intelligence.
♻ ☆ MATCH POLICY: A Simple Pipeline from Point Cloud Registration to Manipulation Policies
Many manipulation tasks require the robot to rearrange objects relative to one another. Such tasks can be described as a sequence of relative poses between parts of a set of rigid bodies. In this work, we propose MATCH POLICY, a simple but novel pipeline for solving high-precision pick and place tasks. Instead of predicting actions directly, our method registers the pick and place targets to the stored demonstrations. This transfers action inference into a point cloud registration task and enables us to realize nontrivial manipulation policies without any training. MATCH POLICY is designed to solve high-precision tasks with a key-frame setting. By leveraging the geometric interaction and the symmetries of the task, it achieves extremely high sample efficiency and generalizability to unseen configurations. We demonstrate its state-of-the-art performance across various tasks on RLBench benchmark compared with several strong baselines and test it on a real robot with six tasks.
comment: project url: https://haojhuang.github.io/match_page/
♻ ☆ Annotation-Free Curb Detection Leveraging Altitude Difference Image
Road curbs are considered as one of the crucial and ubiquitous traffic features, which are essential for ensuring the safety of autonomous vehicles. Current methods for detecting curbs primarily rely on camera imagery or LiDAR point clouds. Image-based methods are vulnerable to fluctuations in lighting conditions and exhibit poor robustness, while methods based on point clouds circumvent the issues associated with lighting variations. However, it is the typical case that significant processing delays are encountered due to the voluminous amount of 3D points contained in each frame of the point cloud data. Furthermore, the inherently unstructured characteristics of point clouds poses challenges for integrating the latest deep learning advancements into point cloud data applications. To address these issues, this work proposes an annotation-free curb detection method leveraging Altitude Difference Image (ADI), which effectively mitigates the aforementioned challenges. Given that methods based on deep learning generally demand extensive, manually annotated datasets, which are both expensive and labor-intensive to create, we present an Automatic Curb Annotator (ACA) module. This module utilizes a deterministic curb detection algorithm to automatically generate a vast quantity of training data. Consequently, it facilitates the training of the curb detection model without necessitating any manual annotation of data. Finally, by incorporating a post-processing module, we manage to achieve state-of-the-art results on the KITTI 3D curb dataset with considerably reduced processing delays compared to existing methods, which underscores the effectiveness of our approach in curb detection tasks.
♻ ☆ Text-driven Adaptation of Foundation Models for Few-shot Surgical Workflow Analysis
Purpose: Surgical workflow analysis is crucial for improving surgical efficiency and safety. However, previous studies rely heavily on large-scale annotated datasets, posing challenges in cost, scalability, and reliance on expert annotations. To address this, we propose Surg-FTDA (Few-shot Text-driven Adaptation), designed to handle various surgical workflow analysis tasks with minimal paired image-label data. Methods: Our approach has two key components. First, Few-shot selection-based modality alignment selects a small subset of images and aligns their embeddings with text embeddings from the downstream task, bridging the modality gap. Second, Text-driven adaptation leverages only text data to train a decoder, eliminating the need for paired image-text data. This decoder is then applied to aligned image embeddings, enabling image-related tasks without explicit image-text pairs. Results: We evaluate our approach to generative tasks (image captioning) and discriminative tasks (triplet recognition and phase recognition). Results show that Surg-FTDA outperforms baselines and generalizes well across downstream tasks. Conclusion: We propose a text-driven adaptation approach that mitigates the modality gap and handles multiple downstream tasks in surgical workflow analysis, with minimal reliance on large annotated datasets. The code and dataset will be released in https://github.com/CAMMA-public/Surg-FTDA
♻ ☆ NavRAG: Generating User Demand Instructions for Embodied Navigation through Retrieval-Augmented LLM
Vision-and-Language Navigation (VLN) is an essential skill for embodied agents, allowing them to navigate in 3D environments following natural language instructions. High-performance navigation models require a large amount of training data, the high cost of manually annotating data has seriously hindered this field. Therefore, some previous methods translate trajectory videos into step-by-step instructions for expanding data, but such instructions do not match well with users' communication styles that briefly describe destinations or state specific needs. Moreover, local navigation trajectories overlook global context and high-level task planning. To address these issues, we propose NavRAG, a retrieval-augmented generation (RAG) framework that generates user demand instructions for VLN. NavRAG leverages LLM to build a hierarchical scene description tree for 3D scene understanding from global layout to local details, then simulates various user roles with specific demands to retrieve from the scene tree, generating diverse instructions with LLM. We annotate over 2 million navigation instructions across 861 scenes and evaluate the data quality and navigation performance of trained models.
♻ ☆ Monocular Depth Estimation and Segmentation for Transparent Object with Iterative Semantic and Geometric Fusion ICRA
Transparent object perception is indispensable for numerous robotic tasks. However, accurately segmenting and estimating the depth of transparent objects remain challenging due to complex optical properties. Existing methods primarily delve into only one task using extra inputs or specialized sensors, neglecting the valuable interactions among tasks and the subsequent refinement process, leading to suboptimal and blurry predictions. To address these issues, we propose a monocular framework, which is the first to excel in both segmentation and depth estimation of transparent objects, with only a single-image input. Specifically, we devise a novel semantic and geometric fusion module, effectively integrating the multi-scale information between tasks. In addition, drawing inspiration from human perception of objects, we further incorporate an iterative strategy, which progressively refines initial features for clearer results. Experiments on two challenging synthetic and real-world datasets demonstrate that our model surpasses state-of-the-art monocular, stereo, and multi-view methods by a large margin of about 38.8%-46.2% with only a single RGB input. Codes and models are publicly available at https://github.com/L-J-Yuan/MODEST.
comment: Accepted by ICRA(2025). The code is accessible through: https://github.com/L-J-Yuan/MODEST
♻ ☆ HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts ICLR 2025
Generalized Category Discovery (GCD) is a challenging task in which, given a partially labelled dataset, models must categorize all unlabelled instances, regardless of whether they come from labelled categories or from new ones. In this paper, we challenge a remaining assumption in this task: that all images share the same domain. Specifically, we introduce a new task and method to handle GCD when the unlabelled data also contains images from different domains to the labelled set. Our proposed `HiLo' networks extract High-level semantic and Low-level domain features, before minimizing the mutual information between the representations. Our intuition is that the clusterings based on domain information and semantic information should be independent. We further extend our method with a specialized domain augmentation tailored for the GCD task, as well as a curriculum learning approach. Finally, we construct a benchmark from corrupted fine-grained datasets as well as a large-scale evaluation on DomainNet with real-world domain shifts, reimplementing a number of GCD baselines in this setting. We demonstrate that HiLo outperforms SoTA category discovery models by a large margin on all evaluations.
comment: v2: Accepted as a conference paper at ICLR 2025; Project page: https://github.com/Visual-AI/hilo/
♻ ☆ CtrLoRA: An Extensible and Efficient Framework for Controllable Image Generation ICLR 2025
Recently, large-scale diffusion models have made impressive progress in text-to-image (T2I) generation. To further equip these T2I models with fine-grained spatial control, approaches like ControlNet introduce an extra network that learns to follow a condition image. However, for every single condition type, ControlNet requires independent training on millions of data pairs with hundreds of GPU hours, which is quite expensive and makes it challenging for ordinary users to explore and develop new types of conditions. To address this problem, we propose the CtrLoRA framework, which trains a Base ControlNet to learn the common knowledge of image-to-image generation from multiple base conditions, along with condition-specific LoRAs to capture distinct characteristics of each condition. Utilizing our pretrained Base ControlNet, users can easily adapt it to new conditions, requiring as few as 1,000 data pairs and less than one hour of single-GPU training to obtain satisfactory results in most scenarios. Moreover, our CtrLoRA reduces the learnable parameters by 90% compared to ControlNet, significantly lowering the threshold to distribute and deploy the model weights. Extensive experiments on various types of conditions demonstrate the efficiency and effectiveness of our method. Codes and model weights will be released at https://github.com/xyfJASON/ctrlora.
comment: ICLR 2025. Code: https://github.com/xyfJASON/ctrlora
♻ ☆ Poison-splat: Computation Cost Attack on 3D Gaussian Splatting ICLR 2025
3D Gaussian splatting (3DGS), known for its groundbreaking performance and efficiency, has become a dominant 3D representation and brought progress to many 3D vision tasks. However, in this work, we reveal a significant security vulnerability that has been largely overlooked in 3DGS: the computation cost of training 3DGS could be maliciously tampered by poisoning the input data. By developing an attack named Poison-splat, we reveal a novel attack surface where the adversary can poison the input images to drastically increase the computation memory and time needed for 3DGS training, pushing the algorithm towards its worst computation complexity. In extreme cases, the attack can even consume all allocable memory, leading to a Denial-of-Service (DoS) that disrupts servers, resulting in practical damages to real-world 3DGS service vendors. Such a computation cost attack is achieved by addressing a bi-level optimization problem through three tailored strategies: attack objective approximation, proxy model rendering, and optional constrained optimization. These strategies not only ensure the effectiveness of our attack but also make it difficult to defend with simple defensive measures. We hope the revelation of this novel attack surface can spark attention to this crucial yet overlooked vulnerability of 3DGS systems. Our code is available at https://github.com/jiahaolu97/poison-splat .
comment: Accepted by ICLR 2025 as a spotlight paper
♻ ☆ FLARE: Feed-forward Geometry, Appearance and Camera Estimation from Uncalibrated Sparse Views CVPR 2025
We present FLARE, a feed-forward model designed to infer high-quality camera poses and 3D geometry from uncalibrated sparse-view images (i.e., as few as 2-8 inputs), which is a challenging yet practical setting in real-world applications. Our solution features a cascaded learning paradigm with camera pose serving as the critical bridge, recognizing its essential role in mapping 3D structures onto 2D image planes. Concretely, FLARE starts with camera pose estimation, whose results condition the subsequent learning of geometric structure and appearance, optimized through the objectives of geometry reconstruction and novel-view synthesis. Utilizing large-scale public datasets for training, our method delivers state-of-the-art performance in the tasks of pose estimation, geometry reconstruction, and novel view synthesis, while maintaining the inference efficiency (i.e., less than 0.5 seconds). The project page and code can be found at: https://zhanghe3z.github.io/FLARE/
comment: CVPR 2025. Website: https://zhanghe3z.github.io/FLARE/
♻ ☆ Optimal Brain Apoptosis ICLR 2025
The increasing complexity and parameter count of Convolutional Neural Networks (CNNs) and Transformers pose challenges in terms of computational efficiency and resource demands. Pruning has been identified as an effective strategy to address these challenges by removing redundant elements such as neurons, channels, or connections, thereby enhancing computational efficiency without heavily compromising performance. This paper builds on the foundational work of Optimal Brain Damage (OBD) by advancing the methodology of parameter importance estimation using the Hessian matrix. Unlike previous approaches that rely on approximations, we introduce Optimal Brain Apoptosis (OBA), a novel pruning method that calculates the Hessian-vector product value directly for each parameter. By decomposing the Hessian matrix across network layers and identifying conditions under which inter-layer Hessian submatrices are non-zero, we propose a highly efficient technique for computing the second-order Taylor expansion of parameters. This approach allows for a more precise pruning process, particularly in the context of CNNs and Transformers, as validated in our experiments including VGG19, ResNet32, ResNet50, and ViT-B/16 on CIFAR10, CIFAR100 and Imagenet datasets. Our code is available at https://github.com/NEU-REAL/OBA.
comment: Accepted to ICLR 2025
♻ ☆ Exploring the Effectiveness of Object-Centric Representations in Visual Question Answering: Comparative Insights with Foundation Models ICLR 2025
Object-centric (OC) representations, which model visual scenes as compositions of discrete objects, have the potential to be used in various downstream tasks to achieve systematic compositional generalization and facilitate reasoning. However, these claims have yet to be thoroughly validated empirically. Recently, foundation models have demonstrated unparalleled capabilities across diverse domains, from language to computer vision, positioning them as a potential cornerstone of future research for a wide range of computational tasks. In this paper, we conduct an extensive empirical study on representation learning for downstream Visual Question Answering (VQA), which requires an accurate compositional understanding of the scene. We thoroughly investigate the benefits and trade-offs of OC models and alternative approaches including large pre-trained foundation models on both synthetic and real-world data, ultimately identifying a promising path to leverage the strengths of both paradigms. The extensiveness of our study, encompassing over 600 downstream VQA models and 15 different types of upstream representations, also provides several additional insights that we believe will be of interest to the community at large.
comment: Published at ICLR 2025
♻ ☆ MIGE: A Unified Framework for Multimodal Instruction-Based Image Generation and Editing
Despite significant progress in diffusion-based image generation, subject-driven generation and instruction-based editing remain challenging. Existing methods typically treat them separately, struggling with limited high-quality data and poor generalization. However, both tasks require capturing complex visual variations while maintaining consistency between inputs and outputs. Therefore, we propose MIGE, a unified framework that standardizes task representations using multimodal instructions. It treats subject-driven generation as creation on a blank canvas and instruction-based editing as modification of an existing image, establishing a shared input-output formulation. MIGE introduces a novel multimodal encoder that maps free-form multimodal instructions into a unified vision-language space, integrating visual and semantic features through a feature fusion mechanism. This unification enables joint training of both tasks, providing two key advantages: (1) Cross-Task Enhancement: By leveraging shared visual and semantic representations, joint training improves instruction adherence and visual consistency in both subject-driven generation and instruction-based editing. (2) Generalization: Learning in a unified format facilitates cross-task knowledge transfer, enabling MIGE to generalize to novel compositional tasks, including instruction-based subject-driven editing. Experiments show that MIGE excels in both subject-driven generation and instruction-based editing while setting a state-of-the-art in the new task of instruction-based subject-driven editing. Code and model have been publicly available at https://github.com/Eureka-Maggie/MIGE.
♻ ☆ Adaptive Prompt: Unlocking the Power of Visual Prompt Tuning
Visual Prompt Tuning (VPT) has recently emerged as a powerful method for adapting pre-trained vision models to downstream tasks. By introducing learnable prompt tokens as task-specific instructions, VPT effectively guides pre-trained transformer models with minimal overhead. Despite its empirical success, a comprehensive theoretical understanding of VPT remains an active area of research. Building on recent insights into the connection between mixture of experts and prompt-based approaches, we identify a key limitation in VPT: the restricted functional expressiveness in prompt formulation. To address this limitation, we propose Visual Adaptive Prompt Tuning (VAPT), a new generation of prompts that redefines prompts as adaptive functions of the input. Our theoretical analysis shows that this simple yet intuitive approach achieves optimal sample efficiency. Empirical results on VTAB-1K and FGVC further demonstrate VAPT's effectiveness, with performance gains of 7.34% and 1.04% over fully fine-tuning baselines, respectively. Notably, VAPT also surpasses VPT by a substantial margin while using fewer parameters. These results highlight both the effectiveness and efficiency of our method and pave the way for future research to explore the potential of adaptive prompts.
comment: 57 pages, 10 figures, 18 tables
♻ ☆ PnP-Flow: Plug-and-Play Image Restoration with Flow Matching
In this paper, we introduce Plug-and-Play (PnP) Flow Matching, an algorithm for solving imaging inverse problems. PnP methods leverage the strength of pre-trained denoisers, often deep neural networks, by integrating them in optimization schemes. While they achieve state-of-the-art performance on various inverse problems in imaging, PnP approaches face inherent limitations on more generative tasks like inpainting. On the other hand, generative models such as Flow Matching pushed the boundary in image sampling yet lack a clear method for efficient use in image restoration. We propose to combine the PnP framework with Flow Matching (FM) by defining a time-dependent denoiser using a pre-trained FM model. Our algorithm alternates between gradient descent steps on the data-fidelity term, reprojections onto the learned FM path, and denoising. Notably, our method is computationally efficient and memory-friendly, as it avoids backpropagation through ODEs and trace computations. We evaluate its performance on denoising, super-resolution, deblurring, and inpainting tasks, demonstrating superior results compared to existing PnP algorithms and Flow Matching based state-of-the-art methods.
♻ ☆ Meta Curvature-Aware Minimization for Domain Generalization
Domain generalization (DG) aims to enhance the ability of models trained on source domains to generalize effectively to unseen domains. Recently, Sharpness-Aware Minimization (SAM) has shown promise in this area by reducing the sharpness of the loss landscape to obtain more generalized models. However, SAM and its variants sometimes fail to guide the model toward a flat minimum, and their training processes exhibit limitations, hindering further improvements in model generalization. In this paper, we first propose an improved model training process aimed at encouraging the model to converge to a flat minima. To achieve this, we design a curvature metric that has a minimal effect when the model is far from convergence but becomes increasingly influential in indicating the curvature of the minima as the model approaches a local minimum. Then we derive a novel algorithm from this metric, called Meta Curvature-Aware Minimization (MeCAM), to minimize the curvature around the local minima. Specifically, the optimization objective of MeCAM simultaneously minimizes the regular training loss, the surrogate gap of SAM, and the surrogate gap of meta-learning. We provide theoretical analysis on MeCAM's generalization error and convergence rate, and demonstrate its superiority over existing DG methods through extensive experiments on five benchmark DG datasets, including PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet. Code will be available on GitHub.
comment: 22 pages, 5 figures, 17 tables
♻ ☆ Towards Training One-Step Diffusion Models Without Distillation
Recent advances in one-step generative models typically follow a two-stage process: first training a teacher diffusion model and then distilling it into a one-step student model. This distillation process traditionally relies on both the teacher model's score function to compute the distillation loss and its weights for student initialization. In this paper, we explore whether one-step generative models can be trained directly without this distillation process. First, we show that the teacher's score function is not essential and propose a family of distillation methods that achieve competitive results without relying on score estimation. Next, we demonstrate that initialization from teacher weights is indispensable in successful training. Surprisingly, we find that this benefit is not due to improved ``input-output" mapping but rather the learned feature representations, which dominate distillation quality. Our findings provide a better understanding of the role of initialization in one-step model training and its impact on distillation quality.
comment: 13 pages, Technical Report
♻ ☆ Foundation Models -- A Panacea for Artificial Intelligence in Pathology?
The role of artificial intelligence (AI) in pathology has evolved from aiding diagnostics to uncovering predictive morphological patterns in whole slide images (WSIs). Recently, foundation models (FMs) leveraging self-supervised pre-training have been widely advocated as a universal solution for diverse downstream tasks. However, open questions remain about their clinical applicability and generalization advantages over end-to-end learning using task-specific (TS) models. Here, we focused on AI with clinical-grade performance for prostate cancer diagnosis and Gleason grading. We present the largest validation of AI for this task, using over 100,000 core needle biopsies from 7,342 patients across 15 sites in 11 countries. We compared two FMs with a fully end-to-end TS model in a multiple instance learning framework. Our findings challenge assumptions that FMs universally outperform TS models. While FMs demonstrated utility in data-scarce scenarios, their performance converged with - and was in some cases surpassed by - TS models when sufficient labeled training data were available. Notably, extensive task-specific training markedly reduced clinically significant misgrading, misdiagnosis of challenging morphologies, and variability across different WSI scanners. Additionally, FMs used up to 35 times more energy than the TS model, raising concerns about their sustainability. Our results underscore that while FMs offer clear advantages for rapid prototyping and research, their role as a universal solution for clinically applicable medical AI remains uncertain. For high-stakes clinical applications, rigorous validation and consideration of task-specific training remain critically important. We advocate for integrating the strengths of FMs and end-to-end learning to achieve robust and resource-efficient AI pathology solutions fit for clinical use.
comment: 50 pages, 15 figures and an appendix (study protocol) which is previously published, see https://doi.org/10.1101/2024.07.04.24309948; updated authors list format
♻ ☆ The PanAf-FGBG Dataset: Understanding the Impact of Backgrounds in Wildlife Behaviour Recognition
Computer vision analysis of camera trap video footage is essential for wildlife conservation, as captured behaviours offer some of the earliest indicators of changes in population health. Recently, several high-impact animal behaviour datasets and methods have been introduced to encourage their use; however, the role of behaviour-correlated background information and its significant effect on out-of-distribution generalisation remain unexplored. In response, we present the PanAf-FGBG dataset, featuring 20 hours of wild chimpanzee behaviours, recorded at over 350 individual camera locations. Uniquely, it pairs every video with a chimpanzee (referred to as a foreground video) with a corresponding background video (with no chimpanzee) from the same camera location. We present two views of the dataset: one with overlapping camera locations and one with disjoint locations. This setup enables, for the first time, direct evaluation of in-distribution and out-of-distribution conditions, and for the impact of backgrounds on behaviour recognition models to be quantified. All clips come with rich behavioural annotations and metadata including unique camera IDs and detailed textual scene descriptions. Additionally, we establish several baselines and present a highly effective latent-space normalisation technique that boosts out-of-distribution performance by +5.42% mAP for convolutional and +3.75% mAP for transformer-based models. Finally, we provide an in-depth analysis on the role of backgrounds in out-of-distribution behaviour recognition, including the so far unexplored impact of background durations (i.e., the count of background frames within foreground videos).
comment: Accepted at the IEEE / CVF Computer Vision and Pattern Recognition Conference 2025
♻ ☆ TRACE: Temporal Grounding Video LLM via Causal Event Modeling ICLR 2025
Video Temporal Grounding (VTG) is a crucial capability for video understanding models and plays a vital role in downstream tasks such as video browsing and editing. To effectively handle various tasks simultaneously and enable zero-shot prediction, there is a growing trend in employing video LLMs for VTG tasks. However, current video LLM-based methods rely exclusively on natural language generation, lacking the ability to model the clear structure inherent in videos, which restricts their effectiveness in tackling VTG tasks. To address this issue, this paper first formally introduces causal event modeling framework, which represents video LLM outputs as sequences of events, and predict the current event using previous events, video inputs, and textural instructions. Each event consists of three components: timestamps, salient scores, and textual captions. We then propose a novel task-interleaved video LLM called TRACE to effectively implement the causal event modeling framework in practice. The TRACE process visual frames, timestamps, salient scores, and text as distinct tasks, employing various encoders and decoding heads for each. Task tokens are arranged in an interleaved sequence according to the causal event modeling framework's formulation. Extensive experiments on various VTG tasks and datasets demonstrate the superior performance of TRACE compared to state-of-the-art video LLMs. Our model and code are available at https://github.com/gyxxyg/TRACE.
comment: ICLR 2025
♻ ☆ Slowing Down Forgetting in Continual Learning
A common challenge in continual learning (CL) is catastrophic forgetting, where the performance on old tasks drops after new, additional tasks are learned. In this paper, we propose a novel framework called ReCL to slow down forgetting in CL. Our framework exploits an implicit bias of gradient-based neural networks due to which these converge to margin maximization points. Such convergence points allow us to reconstruct old data from previous tasks, which we then combine with the current training data. Our framework is flexible and can be applied on top of existing, state-of-the-art CL methods. We further demonstrate the performance gain from our framework across a large series of experiments, including two challenging CL scenarios (class incremental and domain incremental learning), different datasets (MNIST, CIFAR10, TinyImagenet), and different network architectures. Across all experiments, we find large performance gains through ReCL. To the best of our knowledge, our framework is the first to address catastrophic forgetting by leveraging models in CL as their own memory buffers.
♻ ☆ Improving Representation of High-frequency Components for Medical Visual Foundation Models
Foundation models have recently attracted significant attention for their impressive generalizability across diverse downstream tasks. However, these models are demonstrated to exhibit great limitations in representing high-frequency components and fine-grained details. In many medical imaging tasks, the precise representation of such information is crucial due to the inherently intricate anatomical structures, sub-visual features, and complex boundaries involved. Consequently, the limited representation of prevalent foundation models can result in significant performance degradation or even failure in these tasks. To address these challenges, we propose a novel pretraining strategy, named Frequency-advanced Representation Autoencoder (Frepa). Through high-frequency masking and low-frequency perturbation combined with adversarial learning, Frepa encourages the encoder to effectively represent and preserve high-frequency components in the image embeddings. Additionally, we introduce an innovative histogram-equalized image masking strategy, extending the Masked Autoencoder approach beyond ViT to other architectures such as Swin Transformer and convolutional networks. We develop Frepa across nine medical modalities and validate it on 32 downstream tasks for both 2D images and 3D volume data. Without fine-tuning, Frepa can outperform other self-supervised pretraining methods and, in some cases, even surpasses task-specific trained models. This improvement is particularly significant for tasks involving fine-grained details, such as achieving up to a +15% increase in DSC for retina vessel segmentation and a +7% increase in IoU for lung nodule detection. Further experiments quantitatively reveal that Frepa enables superior high-frequency representations and preservation in the embeddings, underscoring its potential for developing more generalized and universal medical image foundation models.
♻ ☆ EXACFS -- A CIL Method to mitigate Catastrophic Forgetting
Deep neural networks (DNNS) excel at learning from static datasets but struggle with continual learning, where data arrives sequentially. Catastrophic forgetting, the phenomenon of forgetting previously learned knowledge, is a primary challenge. This paper introduces EXponentially Averaged Class-wise Feature Significance (EXACFS) to mitigate this issue in the class incremental learning (CIL) setting. By estimating the significance of model features for each learned class using loss gradients, gradually aging the significance through the incremental tasks and preserving the significant features through a distillation loss, EXACFS effectively balances remembering old knowledge (stability) and learning new knowledge (plasticity). Extensive experiments on CIFAR-100 and ImageNet-100 demonstrate EXACFS's superior performance in preserving stability while acquiring plasticity.
♻ ☆ HiBug2: Efficient and Interpretable Error Slice Discovery for Comprehensive Model Debugging
Despite the significant success of deep learning models in computer vision, they often exhibit systematic failures on specific data subsets, known as error slices. Identifying and mitigating these error slices is crucial to enhancing model robustness and reliability in real-world scenarios. In this paper, we introduce HiBug2, an automated framework for error slice discovery and model repair. HiBug2 first generates task-specific visual attributes to highlight instances prone to errors through an interpretable and structured process. It then employs an efficient slice enumeration algorithm to systematically identify error slices, overcoming the combinatorial challenges that arise during slice exploration. Additionally, HiBug2 extends its capabilities by predicting error slices beyond the validation set, addressing a key limitation of prior approaches. Extensive experiments across multiple domains, including image classification, pose estimation, and object detection - show that HiBug2 not only improves the coherence and precision of identified error slices but also significantly enhances the model repair capabilities.
♻ ☆ VoCo-LLaMA: Towards Vision Compression with Large Language Models
Vision-Language Models (VLMs) have achieved remarkable success in various multi-modal tasks, but they are often bottlenecked by the limited context window and high computational cost of processing high-resolution image inputs and videos. Vision compression can alleviate this problem by reducing the vision token count. Previous approaches compress vision tokens with external modules and force LLMs to understand the compressed ones, leading to visual information loss. However, the LLMs' understanding paradigm of vision tokens is not fully utilised in the compression learning process. We propose VoCo-LLaMA, the first approach to compress vision tokens using LLMs. By introducing Vision Compression tokens during the vision instruction tuning phase and leveraging attention distillation, our method distill how LLMs comprehend vision tokens into their processing of VoCo tokens. VoCo-LLaMA facilitates effective vision compression and improves the computational efficiency during the inference stage. Specifically, our method achieves minimal performance loss with a compression ratio of 576$\times$, resulting in up to 94.8$\%$ fewer FLOPs and 69.6$\%$ acceleration in inference time. Furthermore, through continuous training using time-series compressed token sequences of video frames, VoCo-LLaMA demonstrates the ability to understand temporal correlations, outperforming previous methods on popular video question-answering benchmarks. Our approach presents a promising way to unlock the full potential of VLMs' contextual window, enabling more scalable multi-modal applications. The project page, along with the associated code, can be accessed via https://yxxxb.github.io/VoCo-LLaMA-page/.
comment: 11 pages, 4 figures
♻ ☆ Fast and Accurate Gigapixel Pathological Image Classification with Hierarchical Distillation Multi-Instance Learning CVPR2025
Although multi-instance learning (MIL) has succeeded in pathological image classification, it faces the challenge of high inference costs due to processing numerous patches from gigapixel whole slide images (WSIs). To address this, we propose HDMIL, a hierarchical distillation multi-instance learning framework that achieves fast and accurate classification by eliminating irrelevant patches. HDMIL consists of two key components: the dynamic multi-instance network (DMIN) and the lightweight instance pre-screening network (LIPN). DMIN operates on high-resolution WSIs, while LIPN operates on the corresponding low-resolution counterparts. During training, DMIN are trained for WSI classification while generating attention-score-based masks that indicate irrelevant patches. These masks then guide the training of LIPN to predict the relevance of each low-resolution patch. During testing, LIPN first determines the useful regions within low-resolution WSIs, which indirectly enables us to eliminate irrelevant regions in high-resolution WSIs, thereby reducing inference time without causing performance degradation. In addition, we further design the first Chebyshev-polynomials-based Kolmogorov-Arnold classifier in computational pathology, which enhances the performance of HDMIL through learnable activation layers. Extensive experiments on three public datasets demonstrate that HDMIL outperforms previous state-of-the-art methods, e.g., achieving improvements of 3.13% in AUC while reducing inference time by 28.6% on the Camelyon16 dataset.
comment: 11 pages, 4 figures, accepted by CVPR2025
♻ ☆ GDTS: Goal-Guided Diffusion Model with Tree Sampling for Multi-Modal Pedestrian Trajectory Prediction
Accurate prediction of pedestrian trajectories is crucial for improving the safety of autonomous driving. However, this task is generally nontrivial due to the inherent stochasticity of human motion, which naturally requires the predictor to generate multi-modal prediction. Previous works leverage various generative methods, such as GAN and VAE, for pedestrian trajectory prediction. Nevertheless, these methods may suffer from mode collapse and relatively low-quality results. The denoising diffusion probabilistic model (DDPM) has recently been applied to trajectory prediction due to its simple training process and powerful reconstruction ability. However, current diffusion-based methods do not fully utilize input information and usually require many denoising iterations that lead to a long inference time or an additional network for initialization. To address these challenges and facilitate the use of diffusion models in multi-modal trajectory prediction, we propose GDTS, a novel Goal-Guided Diffusion Model with Tree Sampling for multi-modal trajectory prediction. Considering the "goal-driven" characteristics of human motion, GDTS leverages goal estimation to guide the generation of the diffusion network. A two-stage tree sampling algorithm is presented, which leverages common features to reduce the inference time and improve accuracy for multi-modal prediction. Experimental results demonstrate that our proposed framework achieves comparable state-of-the-art performance with real-time inference speed in public datasets.
♻ ☆ CromSS: Cross-modal pre-training with noisy labels for remote sensing image segmentation ICLR 2024
We explore the potential of large-scale noisily labeled data to enhance feature learning by pretraining semantic segmentation models within a multi-modal framework for geospatial applications. We propose a novel Cross-modal Sample Selection (CromSS) method, a weakly supervised pretraining strategy designed to improve feature representations through cross-modal consistency and noise mitigation techniques. Unlike conventional pretraining approaches, CromSS exploits massive amounts of noisy and easy-to-come-by labels for improved feature learning beneficial to semantic segmentation tasks. We investigate middle and late fusion strategies to optimize the multi-modal pretraining architecture design. We also introduce a cross-modal sample selection module to mitigate the adverse effects of label noise, which employs a cross-modal entangling strategy to refine the estimated confidence masks within each modality to guide the sampling process. Additionally, we introduce a spatial-temporal label smoothing technique to counteract overconfidence for enhanced robustness against noisy labels. To validate our approach, we assembled the multi-modal dataset, NoLDO-S12, which consists of a large-scale noisy label subset from Google's Dynamic World (DW) dataset for pretraining and two downstream subsets with high-quality labels from Google DW and OpenStreetMap (OSM) for transfer learning. Experimental results on two downstream tasks and the publicly available DFC2020 dataset demonstrate that when effectively utilized, the low-cost noisy labels can significantly enhance feature learning for segmentation tasks. All data, code, and pretrained weights will be made publicly available.
comment: The 1st short version was accepted as an oral presentation by ICLR 2024 ML4RS workshop. The 2nd extended version is being under review
♻ ☆ Doracamom: Joint 3D Detection and Occupancy Prediction with Multi-view 4D Radars and Cameras for Omnidirectional Perception
3D object detection and occupancy prediction are critical tasks in autonomous driving, attracting significant attention. Despite the potential of recent vision-based methods, they encounter challenges under adverse conditions. Thus, integrating cameras with next-generation 4D imaging radar to achieve unified multi-task perception is highly significant, though research in this domain remains limited. In this paper, we propose Doracamom, the first framework that fuses multi-view cameras and 4D radar for joint 3D object detection and semantic occupancy prediction, enabling comprehensive environmental perception. Specifically, we introduce a novel Coarse Voxel Queries Generator that integrates geometric priors from 4D radar with semantic features from images to initialize voxel queries, establishing a robust foundation for subsequent Transformer-based refinement. To leverage temporal information, we design a Dual-Branch Temporal Encoder that processes multi-modal temporal features in parallel across BEV and voxel spaces, enabling comprehensive spatio-temporal representation learning. Furthermore, we propose a Cross-Modal BEV-Voxel Fusion module that adaptively fuses complementary features through attention mechanisms while employing auxiliary tasks to enhance feature quality. Extensive experiments on the OmniHD-Scenes, View-of-Delft (VoD), and TJ4DRadSet datasets demonstrate that Doracamom achieves state-of-the-art performance in both tasks, establishing new benchmarks for multi-modal 3D perception. Code and models will be publicly available.
♻ ☆ ADUGS-VINS: Generalized Visual-Inertial Odometry for Robust Navigation in Highly Dynamic and Complex Environments
Visual-inertial odometry (VIO) is widely used in various fields, such as robots, drones, and autonomous vehicles. However, real-world scenes often feature dynamic objects, compromising the accuracy of VIO. The diversity and partial occlusion of these objects present a tough challenge for existing dynamic VIO methods. To tackle this challenge, we introduce ADUGS-VINS, which integrates an enhanced SORT algorithm along with a promptable foundation model into VIO, thereby improving pose estimation accuracy in environments with diverse dynamic objects and frequent occlusions. We evaluated our proposed method using multiple public datasets representing various scenes, as well as in a real-world scenario involving diverse dynamic objects. The experimental results demonstrate that our proposed method performs impressively in multiple scenarios, outperforming other state-of-the-art methods. This highlights its remarkable generalization and adaptability in diverse dynamic environments, showcasing its potential to handle various dynamic objects in practical applications.
♻ ☆ Locality-aware Gaussian Compression for Fast and High-quality Rendering ICLR 2025
We present LocoGS, a locality-aware 3D Gaussian Splatting (3DGS) framework that exploits the spatial coherence of 3D Gaussians for compact modeling of volumetric scenes. To this end, we first analyze the local coherence of 3D Gaussian attributes, and propose a novel locality-aware 3D Gaussian representation that effectively encodes locally-coherent Gaussian attributes using a neural field representation with a minimal storage requirement. On top of the novel representation, LocoGS is carefully designed with additional components such as dense initialization, an adaptive spherical harmonics bandwidth scheme and different encoding schemes for different Gaussian attributes to maximize compression performance. Experimental results demonstrate that our approach outperforms the rendering quality of existing compact Gaussian representations for representative real-world 3D datasets while achieving from 54.6$\times$ to 96.6$\times$ compressed storage size and from 2.1$\times$ to 2.4$\times$ rendering speed than 3DGS. Even our approach also demonstrates an averaged 2.4$\times$ higher rendering speed than the state-of-the-art compression method with comparable compression performance.
comment: Accepted to ICLR 2025. Project page: https://seungjooshin.github.io/LocoGS
♻ ☆ RALAD: Bridging the Real-to-Sim Domain Gap in Autonomous Driving with Retrieval-Augmented Learning
In the pursuit of robust autonomous driving systems, models trained on real-world datasets often struggle to adapt to new environments, particularly when confronted with corner cases such as extreme weather conditions. Collecting these corner cases in the real world is non-trivial, which necessitates the use of simulators for validation. However,the high computational cost and the domain gap in data distribution have hindered the seamless transition between real and simulated driving scenarios. To tackle this challenge, we propose Retrieval-Augmented Learning for Autonomous Driving (RALAD), a novel framework designed to bridge the real-to-sim gap at a low cost. RALAD features three primary designs, including (1) domain adaptation via an enhanced Optimal Transport (OT) method that accounts for both individual and grouped image distances, (2) a simple and unified framework that can be applied to various models, and (3) efficient fine-tuning techniques that freeze the computationally expensive layers while maintaining robustness. Experimental results demonstrate that RALAD compensates for the performance degradation in simulated environments while maintaining accuracy in real-world scenarios across three different models. Taking Cross View as an example, the mIOU and mAP metrics in real-world scenarios remain stable before and after RALAD fine-tuning, while in simulated environments,the mIOU and mAP metrics are improved by 10.30% and 12.29%, respectively. Moreover, the re-training cost of our approach is reduced by approximately 88.1%. Our code is available at https://github.com/JiachengZuo/RALAD.git.
♻ ☆ Cross-Spectral Vision Transformer for Biometric Authentication using Forehead Subcutaneous Vein Pattern and Periocular Pattern
Traditional biometric systems have encountered significant setbacks due to various unavoidable factors, for example, face recognition-based biometrics fails due to the wearing of face masks and fingerprints create hygiene concerns. This paper proposes a novel lightweight cross-spectral vision transformer (CS-ViT) for biometric authentication using forehead subcutaneous vein patterns and periocular patterns, offering a promising alternative to traditional methods, capable of performing well even with the face masks and without any physical touch. The proposed framework comprises a cross-spectral dual-channel architecture designed to handle two distinct biometric traits and to capture inter-dependencies in terms of relative spectral patterns. Each channel consists of a Phase-Only Correlation Cross-Spectral Attention (POC-CSA) that captures their individual as well as correlated patterns. The computation of cross-spectral attention using POC extracts the phase correlation in the spatial features. Therefore, it is robust against the resolution/intensity variations and illumination of the input images, assuming both biometric traits are from the same person. The lightweight model is suitable for edge device deployment. The performance of the proposed algorithm was rigorously evaluated using the Forehead Subcutaneous Vein Pattern and Periocular Biometric Pattern (FSVP-PBP) database. The results demonstrated the superiority of the algorithm over state-of-the-art methods, achieving a remarkable classification accuracy of 98.8% with the combined vein and periocular patterns.
comment: Submitted to IEEE TPAMI
♻ ☆ 3D-AffordanceLLM: Harnessing Large Language Models for Open-Vocabulary Affordance Detection in 3D Worlds ICLR
3D Affordance detection is a challenging problem with broad applications on various robotic tasks. Existing methods typically formulate the detection paradigm as a label-based semantic segmentation task. This paradigm relies on predefined labels and lacks the ability to comprehend complex natural language, resulting in limited generalization in open-world scene. To address these limitations, we reformulate the traditional affordance detection paradigm into \textit{Instruction Reasoning Affordance Segmentation} (IRAS) task. This task is designed to output a affordance mask region given a query reasoning text, which avoids fixed categories of input labels. We accordingly propose the \textit{3D-AffordanceLLM} (3D-ADLLM), a framework designed for reasoning affordance detection in 3D open-scene. Specifically, 3D-ADLLM introduces large language models (LLMs) to 3D affordance perception with a custom-designed decoder for generating affordance masks, thus achieving open-world reasoning affordance detection. In addition, given the scarcity of 3D affordance datasets for training large models, we seek to extract knowledge from general segmentation data and transfer it to affordance detection. Thus, we propose a multi-stage training strategy that begins with a novel pre-training task, i.e., \textit{Referring Object Part Segmentation}~(ROPS). This stage is designed to equip the model with general recognition and segmentation capabilities at the object-part level. Then followed by fine-tuning with the IRAS task, 3D-ADLLM obtains the reasoning ability for affordance detection. In summary, 3D-ADLLM leverages the rich world knowledge and human-object interaction reasoning ability of LLMs, achieving approximately an 8\% improvement in mIoU on open-vocabulary affordance detection tasks.
comment: ICLR
♻ ☆ Representation Engineering: A Top-Down Approach to AI Transparency
In this paper, we identify and characterize the emerging area of representation engineering (RepE), an approach to enhancing the transparency of AI systems that draws on insights from cognitive neuroscience. RepE places population-level representations, rather than neurons or circuits, at the center of analysis, equipping us with novel methods for monitoring and manipulating high-level cognitive phenomena in deep neural networks (DNNs). We provide baselines and an initial analysis of RepE techniques, showing that they offer simple yet effective solutions for improving our understanding and control of large language models. We showcase how these methods can provide traction on a wide range of safety-relevant problems, including honesty, harmlessness, power-seeking, and more, demonstrating the promise of top-down transparency research. We hope that this work catalyzes further exploration of RepE and fosters advancements in the transparency and safety of AI systems.
comment: Code is available at https://github.com/andyzoujm/representation-engineering
♻ ☆ Low-Biased General Annotated Dataset Generation
Pre-training backbone networks on a general annotated dataset (e.g., ImageNet) that comprises numerous manually collected images with category annotations has proven to be indispensable for enhancing the generalization capacity of downstream visual tasks. However, those manually collected images often exhibit bias, which is non-transferable across either categories or domains, thus causing the model's generalization capacity degeneration. To mitigate this problem, we present an low-biased general annotated dataset generation framework (lbGen). Instead of expensive manual collection, we aim at directly generating low-biased images with category annotations. To achieve this goal, we propose to leverage the advantage of a multimodal foundation model (e.g., CLIP), in terms of aligning images in an low-biased semantic space defined by language. Specifically, we develop a bi-level semantic alignment loss, which not only forces all generated images to be consistent with the semantic distribution of all categories belonging to the target dataset in an adversarial learning manner, but also requires each generated image to match the semantic description of its category name. In addition, we further cast an existing image quality scoring model into a quality assurance loss to preserve the quality of the generated image. By leveraging these two loss functions, we can obtain an low-biased image generation model by simply fine-tuning a pre-trained diffusion model using only all category names in the target dataset as input. Experimental results confirm that, compared with the manually labeled dataset or other synthetic datasets, the utilization of our generated low-biased datasets leads to stable generalization capacity enhancement of different backbone networks across various tasks, especially in tasks where the manually labeled samples are scarce.
comment: Preprint
♻ ☆ Structural-Entropy-Based Sample Selection for Efficient and Effective Learning ICLR 2025
Sample selection improves the efficiency and effectiveness of machine learning models by providing informative and representative samples. Typically, samples can be modeled as a sample graph, where nodes are samples and edges represent their similarities. Most existing methods are based on local information, such as the training difficulty of samples, thereby overlooking global information, such as connectivity patterns. This oversight can result in suboptimal selection because global information is crucial for ensuring that the selected samples well represent the structural properties of the graph. To address this issue, we employ structural entropy to quantify global information and losslessly decompose it from the whole graph to individual nodes using the Shapley value. Based on the decomposition, we present $\textbf{S}$tructural-$\textbf{E}$ntropy-based sample $\textbf{S}$election ($\textbf{SES}$), a method that integrates both global and local information to select informative and representative samples. SES begins by constructing a $k$NN-graph among samples based on their similarities. It then measures sample importance by combining structural entropy (global metric) with training difficulty (local metric). Finally, SES applies importance-biased blue noise sampling to select a set of diverse and representative samples. Comprehensive experiments on three learning scenarios -- supervised learning, active learning, and continual learning -- clearly demonstrate the effectiveness of our method.
comment: Published as a conference paper at ICLR 2025
♻ ☆ Dynamic Gaussians Mesh: Consistent Mesh Reconstruction from Dynamic Scenes
Modern 3D engines and graphics pipelines require mesh as a memory-efficient representation, which allows efficient rendering, geometry processing, texture editing, and many other downstream operations. However, it is still highly difficult to obtain high-quality mesh in terms of detailed structure and time consistency from dynamic observations. To this end, we introduce Dynamic Gaussians Mesh (DG-Mesh), a framework to reconstruct a high-fidelity and time-consistent mesh from dynamic input. Our work leverages the recent advancement in 3D Gaussian Splatting to construct the mesh sequence with temporal consistency from dynamic observations. Building on top of this representation, DG-Mesh recovers high-quality meshes from the Gaussian points and can track the mesh vertices over time, which enables applications such as texture editing on dynamic objects. We introduce the Gaussian-Mesh Anchoring, which encourages evenly distributed Gaussians, resulting better mesh reconstruction through mesh-guided densification and pruning on the deformed Gaussians. By applying cycle-consistent deformation between the canonical and the deformed space, we can project the anchored Gaussian back to the canonical space and optimize Gaussians across all time frames. During the evaluation on different datasets, DG-Mesh provides significantly better mesh reconstruction and rendering than baselines. Project page: https://www.liuisabella.com/DG-Mesh
comment: Project page: https://www.liuisabella.com/DG-Mesh
♻ ☆ ESVO2: Direct Visual-Inertial Odometry with Stereo Event Cameras
Event-based visual odometry is a specific branch of visual Simultaneous Localization and Mapping (SLAM) techniques, which aims at solving tracking and mapping subproblems (typically in parallel), by exploiting the special working principles of neuromorphic (i.e., event-based) cameras. Due to the motion-dependent nature of event data, explicit data association (i.e., feature matching) under large-baseline view-point changes is difficult to establish, making direct methods a more rational choice. However, state-of-the-art direct methods are limited by the high computational complexity of the mapping sub-problem and the degeneracy of camera pose tracking in certain degrees of freedom (DoF) in rotation. In this paper, we tackle these issues by building an event-based stereo visual-inertial odometry system on top of a direct pipeline. Specifically, to speed up the mapping operation, we propose an efficient strategy for sampling contour points according to the local dynamics of events. The mapping performance is also improved in terms of structure completeness and local smoothness by merging the temporal stereo and static stereo results. To circumvent the degeneracy of camera pose tracking in recovering the pitch and yaw components of general 6-DoF motion, we introduce IMU measurements as motion priors via pre-integration. To this end, a compact back-end is proposed for continuously updating the IMU bias and predicting the linear velocity, enabling an accurate motion prediction for camera pose tracking. The resulting system scales well with modern high-resolution event cameras and leads to better global positioning accuracy in large-scale outdoor environments. Extensive evaluations on five publicly available datasets featuring different resolutions and scenarios justify the superior performance of the proposed system against five state-of-the-art methods.
♻ ☆ OMG: Opacity Matters in Material Modeling with Gaussian Splatting ICLR 2025
Decomposing geometry, materials and lighting from a set of images, namely inverse rendering, has been a long-standing problem in computer vision and graphics. Recent advances in neural rendering enable photo-realistic and plausible inverse rendering results. The emergence of 3D Gaussian Splatting has boosted it to the next level by showing real-time rendering potentials. An intuitive finding is that the models used for inverse rendering do not take into account the dependency of opacity w.r.t. material properties, namely cross section, as suggested by optics. Therefore, we develop a novel approach that adds this dependency to the modeling itself. Inspired by radiative transfer, we augment the opacity term by introducing a neural network that takes as input material properties to provide modeling of cross section and a physically correct activation function. The gradients for material properties are therefore not only from color but also from opacity, facilitating a constraint for their optimization. Therefore, the proposed method incorporates more accurate physical properties compared to previous works. We implement our method into 3 different baselines that use Gaussian Splatting for inverse rendering and achieve significant improvements universally in terms of novel view synthesis and material modeling.
comment: Published as a conference paper at ICLR 2025
♻ ☆ PATCH: a deep learning method to assess heterogeneity of artistic practice in historical paintings
The history of art has seen significant shifts in the manner in which artworks are created, making understanding of creative processes a central question in technical art history. In the Renaissance and Early Modern period, paintings were largely produced by master painters directing workshops of apprentices who often contributed to projects. The masters varied significantly in artistic and managerial styles, meaning different combinations of artists and implements might be seen both between masters and within workshops or even individual canvases. Information on how different workshops were managed and the processes by which artworks were created remains elusive. Machine learning methods have potential to unearth new information about artists' creative processes by extending the analysis of brushwork to a microscopic scale. Analysis of workshop paintings, however, presents a challenge in that documentation of the artists and materials involved is sparse, meaning external examples are not available to train networks to recognize their contributions. Here we present a novel machine learning approach we call pairwise assignment training for classifying heterogeneity (PATCH) that is capable of identifying individual artistic practice regimes with no external training data, or "ground truth." The method achieves unsupervised results by supervised means, and outperforms both simple statistical procedures and unsupervised machine learning methods. We apply this method to two historical paintings by the Spanish Renaissance master, El Greco: The Baptism of Christ and Christ on the Cross with Landscape, and our findings regarding the former potentially challenge previous work that has assigned the painting to workshop members. Further, the results of our analyses create a measure of heterogeneity of artistic practice that can be used to characterize artworks across time and space.
comment: main text: 16 pages, 6 figures; SI: 7 pages, 3 figures; v2: minor typo corrections, higher resolution figures
♻ ☆ S-NeRF++: Autonomous Driving Simulation via Neural Reconstruction and Generation
Autonomous driving simulation system plays a crucial role in enhancing self-driving data and simulating complex and rare traffic scenarios, ensuring navigation safety. However, traditional simulation systems, which often heavily rely on manual modeling and 2D image editing, struggled with scaling to extensive scenes and generating realistic simulation data. In this study, we present S-NeRF++, an innovative autonomous driving simulation system based on neural reconstruction. Trained on widely-used self-driving datasets such as nuScenes and Waymo, S-NeRF++ can generate a large number of realistic street scenes and foreground objects with high rendering quality as well as offering considerable flexibility in manipulation and simulation. Specifically, S-NeRF++ is an enhanced neural radiance field for synthesizing large-scale scenes and moving vehicles, with improved scene parameterization and camera pose learning. The system effectively utilizes noisy and sparse LiDAR data to refine training and address depth outliers, ensuring high-quality reconstruction and novel-view rendering. It also provides a diverse foreground asset bank by reconstructing and generating different foreground vehicles to support comprehensive scenario creation.Moreover, we have developed an advanced foreground-background fusion pipeline that skillfully integrates illumination and shadow effects, further enhancing the realism of our simulations. With the high-quality simulated data provided by our S-NeRF++, we found the perception methods enjoy performance boosts on several autonomous driving downstream tasks, further demonstrating our proposed simulator's effectiveness.
comment: IEEE TPAMI 2025
♻ ☆ AdvLogo: Adversarial Patch Attack against Object Detectors based on Diffusion Models
With the rapid development of deep learning, object detectors have demonstrated impressive performance; however, vulnerabilities still exist in certain scenarios. Current research exploring the vulnerabilities using adversarial patches often struggles to balance the trade-off between attack effectiveness and visual quality. To address this problem, we propose a novel framework of patch attack from semantic perspective, which we refer to as AdvLogo. Based on the hypothesis that every semantic space contains an adversarial subspace where images can cause detectors to fail in recognizing objects, we leverage the semantic understanding of the diffusion denoising process and drive the process to adversarial subareas by perturbing the latent and unconditional embeddings at the last timestep. To mitigate the distribution shift that exposes a negative impact on image quality, we apply perturbation to the latent in frequency domain with the Fourier Transform. Experimental results demonstrate that AdvLogo achieves strong attack performance while maintaining high visual quality.
♻ ☆ DynamicCity: Large-Scale 4D Occupancy Generation from Dynamic Scenes ICLR 2025
Urban scene generation has been developing rapidly recently. However, existing methods primarily focus on generating static and single-frame scenes, overlooking the inherently dynamic nature of real-world driving environments. In this work, we introduce DynamicCity, a novel 4D occupancy generation framework capable of generating large-scale, high-quality dynamic 4D scenes with semantics. DynamicCity mainly consists of two key models. 1) A VAE model for learning HexPlane as the compact 4D representation. Instead of using naive averaging operations, DynamicCity employs a novel Projection Module to effectively compress 4D features into six 2D feature maps for HexPlane construction, which significantly enhances HexPlane fitting quality (up to 12.56 mIoU gain). Furthermore, we utilize an Expansion & Squeeze Strategy to reconstruct 3D feature volumes in parallel, which improves both network training efficiency and reconstruction accuracy than naively querying each 3D point (up to 7.05 mIoU gain, 2.06x training speedup, and 70.84% memory reduction). 2) A DiT-based diffusion model for HexPlane generation. To make HexPlane feasible for DiT generation, a Padded Rollout Operation is proposed to reorganize all six feature planes of the HexPlane as a squared 2D feature map. In particular, various conditions could be introduced in the diffusion or sampling process, supporting versatile 4D generation applications, such as trajectory- and command-driven generation, inpainting, and layout-conditioned generation. Extensive experiments on the CarlaSC and Waymo datasets demonstrate that DynamicCity significantly outperforms existing state-of-the-art 4D occupancy generation methods across multiple metrics. The code and models have been released to facilitate future research.
comment: ICLR 2025 Spotlight; 35 pages, 18 figures, 15 tables; Project Page at https://dynamic-city.github.io/
♻ ☆ Calib3D: Calibrating Model Preferences for Reliable 3D Scene Understanding WACV 2025
Safety-critical 3D scene understanding tasks necessitate not only accurate but also confident predictions from 3D perception models. This study introduces Calib3D, a pioneering effort to benchmark and scrutinize the reliability of 3D scene understanding models from an uncertainty estimation viewpoint. We comprehensively evaluate 28 state-of-the-art models across 10 diverse 3D datasets, uncovering insightful phenomena that cope with both the aleatoric and epistemic uncertainties in 3D scene understanding. We discover that despite achieving impressive levels of accuracy, existing models frequently fail to provide reliable uncertainty estimates -- a pitfall that critically undermines their applicability in safety-sensitive contexts. Through extensive analysis of key factors such as network capacity, LiDAR representations, rasterization resolutions, and 3D data augmentation techniques, we correlate these aspects directly with the model calibration efficacy. Furthermore, we introduce DeptS, a novel depth-aware scaling approach aimed at enhancing 3D model calibration. Extensive experiments across a wide range of configurations validate the superiority of our method. We hope this work could serve as a cornerstone for fostering reliable 3D scene understanding. Code and benchmark toolkit are publicly available.
comment: WACV 2025 Oral; 26 pages, 8 figures, 12 tables; Code at https://github.com/ldkong1205/Calib3D
♻ ☆ Tuning Timestep-Distilled Diffusion Model Using Pairwise Sample Optimization
Recent advancements in timestep-distilled diffusion models have enabled high-quality image generation that rivals non-distilled multi-step models, but with significantly fewer inference steps. While such models are attractive for applications due to the low inference cost and latency, fine-tuning them with a naive diffusion objective would result in degraded and blurry outputs. An intuitive alternative is to repeat the diffusion distillation process with a fine-tuned teacher model, which produces good results but is cumbersome and computationally intensive; the distillation training usually requires magnitude higher of training compute compared to fine-tuning for specific image styles. In this paper, we present an algorithm named pairwise sample optimization (PSO), which enables the direct fine-tuning of an arbitrary timestep-distilled diffusion model. PSO introduces additional reference images sampled from the current time-step distilled model, and increases the relative likelihood margin between the training images and reference images. This enables the model to retain its few-step generation ability, while allowing for fine-tuning of its output distribution. We also demonstrate that PSO is a generalized formulation which can be flexibly extended to both offline-sampled and online-sampled pairwise data, covering various popular objectives for diffusion model preference optimization. We evaluate PSO in both preference optimization and other fine-tuning tasks, including style transfer and concept customization. We show that PSO can directly adapt distilled models to human-preferred generation with both offline and online-generated pairwise preference image data. PSO also demonstrates effectiveness in style transfer and concept customization by directly tuning timestep-distilled diffusion models.
♻ ☆ FlexDrive: Toward Trajectory Flexibility in Driving Scene Reconstruction and Rendering
Driving scene reconstruction and rendering have advanced significantly using the 3D Gaussian Splatting. However, most prior research has focused on the rendering quality along a pre-recorded vehicle path and struggles to generalize to out-of-path viewpoints, which is caused by the lack of high-quality supervision in those out-of-path views. To address this issue, we introduce an Inverse View Warping technique to create compact and high-quality images as supervision for the reconstruction of the out-of-path views, enabling high-quality rendering results for those views. For accurate and robust inverse view warping, a depth bootstrap strategy is proposed to obtain on-the-fly dense depth maps during the optimization process, overcoming the sparsity and incompleteness of LiDAR depth data. Our method achieves superior in-path and out-of-path reconstruction and rendering performance on the widely used Waymo Open dataset. In addition, a simulator-based benchmark is proposed to obtain the out-of-path ground truth and quantitatively evaluate the performance of out-of-path rendering, where our method outperforms previous methods by a significant margin.
♻ ☆ Semantically Structured Image Compression via Irregular Group-Based Decoupling ICCV2023
Image compression techniques typically focus on compressing rectangular images for human consumption, however, resulting in transmitting redundant content for downstream applications. To overcome this limitation, some previous works propose to semantically structure the bitstream, which can meet specific application requirements by selective transmission and reconstruction. Nevertheless, they divide the input image into multiple rectangular regions according to semantics and ignore avoiding information interaction among them, causing waste of bitrate and distorted reconstruction of region boundaries. In this paper, we propose to decouple an image into multiple groups with irregular shapes based on a customized group mask and compress them independently. Our group mask describes the image at a finer granularity, enabling significant bitrate saving by reducing the transmission of redundant content. Moreover, to ensure the fidelity of selective reconstruction, this paper proposes the concept of group-independent transform that maintain the independence among distinct groups. And we instantiate it by the proposed Group-Independent Swin-Block (GI Swin-Block). Experimental results demonstrate that our framework structures the bitstream with negligible cost, and exhibits superior performance on both visual quality and intelligent task supporting.
comment: Accept by ICCV2023
♻ ☆ Learning to Learn Weight Generation via Trajectory Diffusion
Diffusion-based algorithms have emerged as promising techniques for weight generation, particularly in scenarios like multi-task learning that require frequent weight updates. However, existing solutions suffer from limited cross-task transferability. In addition, they only utilize optimal weights as training samples, ignoring the value of other weights in the optimization process. To address these issues, we propose Lt-Di, which integrates the diffusion algorithm with meta-learning to generate weights for unseen tasks. Furthermore, we extend the vanilla diffusion algorithm into a trajectory diffusion algorithm to utilize other weights along the optimization trajectory. Trajectory diffusion decomposes the entire diffusion chain into multiple shorter ones, improving training and inference efficiency. We analyze the convergence properties of the weight generation paradigm and improve convergence efficiency without additional time overhead. Our experiments demonstrate Lt-Di's higher accuracy while reducing computational overhead across various tasks, including zero-shot and few-shot learning, multi-domain generalization, and large-scale language model fine-tuning.Our code is released at https://anonymous.4open.science/r/Lt-Di-0E51.
Multi-modal AI for comprehensive breast cancer prognostication
Treatment selection in breast cancer is guided by molecular subtypes and clinical characteristics. However, current tools including genomic assays lack the accuracy required for optimal clinical decision-making. We developed a novel artificial intelligence (AI)-based approach that integrates digital pathology images with clinical data, providing a more robust and effective method for predicting the risk of cancer recurrence in breast cancer patients. Specifically, we utilized a vision transformer pan-cancer foundation model trained with self-supervised learning to extract features from digitized H&E-stained slides. These features were integrated with clinical data to form a multi-modal AI test predicting cancer recurrence and death. The test was developed and evaluated using data from a total of 8,161 female breast cancer patients across 15 cohorts originating from seven countries. Of these, 3,502 patients from five cohorts were used exclusively for evaluation, while the remaining patients were used for training. Our test accurately predicted our primary endpoint, disease-free interval, in the five evaluation cohorts (C-index: 0.71 [0.68-0.75], HR: 3.63 [3.02-4.37, p<0.001]). In a direct comparison (n=858), the AI test was more accurate than Oncotype DX, the standard-of-care 21-gene assay, achieving a C-index of 0.67 [0.61-0.74] versus 0.61 [0.49-0.73], respectively. Additionally, the AI test added independent prognostic information to Oncotype DX in a multivariate analysis (HR: 3.11 [1.91-5.09, p<0.001)]). The test demonstrated robust accuracy across major molecular breast cancer subtypes, including TNBC (C-index: 0.71 [0.62-0.81], HR: 3.81 [2.35-6.17, p=0.02]), where no diagnostic tools are currently recommended by clinical guidelines. These results suggest that our AI test improves upon the accuracy of existing prognostic tests, while being applicable to a wider range of patients.
♻ ☆ A Survey on Vision-Language-Action Models for Embodied AI
Embodied AI is widely recognized as a key element of artificial general intelligence because it involves controlling embodied agents to perform tasks in the physical world. Building on the success of large language models and vision-language models, a new category of multimodal models -- referred to as vision-language-action models (VLAs) -- has emerged to address language-conditioned robotic tasks in embodied AI by leveraging their distinct ability to generate actions. In recent years, a myriad of VLAs have been developed, making it imperative to capture the rapidly evolving landscape through a comprehensive survey. To this end, we present the first survey on VLAs for embodied AI. This work provides a detailed taxonomy of VLAs, organized into three major lines of research. The first line focuses on individual components of VLAs. The second line is dedicated to developing control policies adept at predicting low-level actions. The third line comprises high-level task planners capable of decomposing long-horizon tasks into a sequence of subtasks, thereby guiding VLAs to follow more general user instructions. Furthermore, we provide an extensive summary of relevant resources, including datasets, simulators, and benchmarks. Finally, we discuss the challenges faced by VLAs and outline promising future directions in embodied AI.
comment: 16 pages, a survey of vision-language-action models
♻ ☆ GAMED-Snake: Gradient-aware Adaptive Momentum Evolution Deep Snake Model for Multi-organ Segmentation
Multi-organ segmentation is a critical yet challenging task due to complex anatomical backgrounds, blurred boundaries, and diverse morphologies. This study introduces the Gradient-aware Adaptive Momentum Evolution Deep Snake (GAMED-Snake) model, which establishes a novel paradigm for contour-based segmentation by integrating gradient-based learning with adaptive momentum evolution mechanisms. The GAMED-Snake model incorporates three major innovations: First, the Distance Energy Map Prior (DEMP) generates a pixel-level force field that effectively attracts contour points towards the true boundaries, even in scenarios with complex backgrounds and blurred edges. Second, the Differential Convolution Inception Module (DCIM) precisely extracts comprehensive energy gradients, significantly enhancing segmentation accuracy. Third, the Adaptive Momentum Evolution Mechanism (AMEM) employs cross-attention to establish dynamic features across different iterations of evolution, enabling precise boundary alignment for diverse morphologies. Experimental results on four challenging multi-organ segmentation datasets demonstrate that GAMED-Snake improves the mDice metric by approximately 2% compared to state-of-the-art methods. Code will be available at https://github.com/SYSUzrc/GAMED-Snake.
♻ ☆ Noise2Score3D:Unsupervised Tweedie's Approach for Point Cloud Denoising
Building on recent advances in Bayesian statistics and image denoising, we propose Noise2Score3D, a fully unsupervised framework for point cloud denoising that addresses the critical challenge of limited availability of clean data. Noise2Score3D learns the gradient of the underlying point cloud distribution directly from noisy data, eliminating the need for clean data during training. By leveraging Tweedie's formula, our method performs inference in a single step, avoiding the iterative processes used in existing unsupervised methods, thereby improving both performance and efficiency. Experimental results demonstrate that Noise2Score3D achieves state-of-the-art performance on standard benchmarks, outperforming other unsupervised methods in Chamfer distance and point-to-mesh metrics, and rivaling some supervised approaches. Furthermore, Noise2Score3D demonstrates strong generalization ability beyond training datasets. Additionally, we introduce Total Variation for Point Cloud, a criterion that allows for the estimation of unknown noise parameters, which further enhances the method's versatility and real-world utility.
♻ ☆ MMed-RAG: Versatile Multimodal RAG System for Medical Vision Language Models ICLR 2025
Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection method, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in the factual accuracy of Med-LVLMs. Our data and code are available in https://github.com/richard-peng-xia/MMed-RAG.
comment: ICLR 2025
♻ ☆ Towards Generalizable Scene Change Detection CVPR 2025
While current state-of-the-art Scene Change Detection (SCD) approaches achieve impressive results in well-trained research data, they become unreliable under unseen environments and different temporal conditions; in-domain performance drops from 77.6\% to 8.0\% in a previously unseen environment and to 4.6\% under a different temporal condition -- calling for generalizable SCD and benchmark. In this work, we propose the Generalizable Scene Change Detection Framework (GeSCF), which addresses unseen domain performance and temporal consistency -- to meet the growing demand for anything SCD. Our method leverages the pre-trained Segment Anything Model (SAM) in a zero-shot manner. For this, we design Initial Pseudo-mask Generation and Geometric-Semantic Mask Matching -- seamlessly turning user-guided prompt and single-image based segmentation into scene change detection for a pair of inputs without guidance. Furthermore, we define the Generalizable Scene Change Detection (GeSCD) benchmark along with novel metrics and an evaluation protocol to facilitate SCD research in generalizability. In the process, we introduce the ChangeVPR dataset, a collection of challenging image pairs with diverse environmental scenarios -- including urban, suburban, and rural settings. Extensive experiments across various datasets demonstrate that GeSCF achieves an average performance gain of 19.2\% on existing SCD datasets and 30.0\% on the ChangeVPR dataset, nearly doubling the prior art performance. We believe our work can lay a solid foundation for robust and generalizable SCD research.
comment: Manuscript. Accepted to CVPR 2025
♻ ☆ MAA: Meticulous Adversarial Attack against Vision-Language Pre-trained Models
Current adversarial attacks for evaluating the robustness of vision-language pre-trained (VLP) models in multi-modal tasks suffer from limited transferability, where attacks crafted for a specific model often struggle to generalize effectively across different models, limiting their utility in assessing robustness more broadly. This is mainly attributed to the over-reliance on model-specific features and regions, particularly in the image modality. In this paper, we propose an elegant yet highly effective method termed Meticulous Adversarial Attack (MAA) to fully exploit model-independent characteristics and vulnerabilities of individual samples, achieving enhanced generalizability and reduced model dependence. MAA emphasizes fine-grained optimization of adversarial images by developing a novel resizing and sliding crop (RScrop) technique, incorporating a multi-granularity similarity disruption (MGSD) strategy. Extensive experiments across diverse VLP models, multiple benchmark datasets, and a variety of downstream tasks demonstrate that MAA significantly enhances the effectiveness and transferability of adversarial attacks. A large cohort of performance studies is conducted to generate insights into the effectiveness of various model configurations, guiding future advancements in this domain.
♻ ☆ RobotFingerPrint: Unified Gripper Coordinate Space for Multi-Gripper Grasp Synthesis and Transfer
We introduce a novel grasp representation named the Unified Gripper Coordinate Space (UGCS) for grasp synthesis and grasp transfer. Our representation leverages spherical coordinates to create a shared coordinate space across different robot grippers, enabling it to synthesize and transfer grasps for both novel objects and previously unseen grippers. The strength of this representation lies in the ability to map palm and fingers of a gripper and the unified coordinate space. Grasp synthesis is formulated as predicting the unified spherical coordinates on object surface points via a conditional variational autoencoder. The predicted unified gripper coordinates establish exact correspondences between the gripper and object points, which is used to optimize grasp pose and joint values. Grasp transfer is facilitated through the point-to-point correspondence between any two (potentially unseen) grippers and solved via a similar optimization. Extensive simulation and real-world experiments showcase the efficacy of the unified grasp representation for grasp synthesis in generating stable and diverse grasps. Similarly, we showcase real-world grasp transfer from human demonstrations across different objects.
comment: 8 pages, 11 figures, 3 tables. Project page available at https://irvlutd.github.io/RobotFingerPrint
♻ ☆ The Labyrinth of Links: Navigating the Associative Maze of Multi-modal LLMs ICLR 2025
Multi-modal Large Language Models (MLLMs) have exhibited impressive capability. However, recently many deficiencies of MLLMs have been found compared to human intelligence, $\textit{e.g.}$, hallucination. To drive the MLLMs study, the community dedicated efforts to building larger benchmarks with complex tasks. In this paper, we propose benchmarking an essential but usually overlooked intelligence: $\textbf{association}$, a human's basic capability to link observation and prior practice memory. To comprehensively investigate MLLM's performance on the association, we formulate the association task and devise a standard benchmark based on adjective and verb semantic concepts. Instead of costly data annotation and curation, we propose a convenient $\textbf{annotation-free}$ construction method transforming the general dataset for our association tasks. Simultaneously, we devise a rigorous data refinement process to eliminate confusion in the raw dataset. Building on this database, we establish three levels of association tasks: single-step, synchronous, and asynchronous associations. Moreover, we conduct a comprehensive investigation into the MLLMs' zero-shot association capabilities, addressing multiple dimensions, including three distinct memory strategies, both open-source and closed-source MLLMs, cutting-edge Mixture-of-Experts (MoE) models, and the involvement of human experts. Our systematic investigation shows that current open-source MLLMs consistently exhibit poor capability in our association tasks, even the currently state-of-the-art GPT-4V(vision) also has a significant gap compared to humans. We believe our benchmark would pave the way for future MLLM studies. $\textit{Our data and code are available at:}$ https://mvig-rhos.com/llm_inception.
comment: Accepted by ICLR 2025. Project page: https://mvig-rhos.com/llm_inception
♻ ☆ Self-Supervised Contrastive Learning for Videos using Differentiable Local Alignment BMVC
Robust frame-wise embeddings are essential to perform video analysis and understanding tasks. We present a self-supervised method for representation learning based on aligning temporal video sequences. Our framework uses a transformer-based encoder to extract frame-level features and leverages them to find the optimal alignment path between video sequences. We introduce the novel Local-Alignment Contrastive (LAC) loss, which combines a differentiable local alignment loss to capture local temporal dependencies with a contrastive loss to enhance discriminative learning. Prior works on video alignment have focused on using global temporal ordering across sequence pairs, whereas our loss encourages identifying the best-scoring subsequence alignment. LAC uses the differentiable Smith-Waterman (SW) affine method, which features a flexible parameterization learned through the training phase, enabling the model to adjust the temporal gap penalty length dynamically. Evaluations show that our learned representations outperform existing state-of-the-art approaches on action recognition tasks.
comment: Accepted in 2nd Workshop on Video Understanding and its Applications, held in conjunction with the British Machine Vision Conference (BMVC) 2024
♻ ☆ ModeDreamer: Mode Guiding Score Distillation for Text-to-3D Generation using Reference Image Prompts
Existing Score Distillation Sampling (SDS)-based methods have driven significant progress in text-to-3D generation. However, 3D models produced by SDS-based methods tend to exhibit over-smoothing and low-quality outputs. These issues arise from the mode-seeking behavior of current methods, where the scores used to update the model oscillate between multiple modes, resulting in unstable optimization and diminished output quality. To address this problem, we introduce a novel image prompt score distillation loss named ISD, which employs a reference image to direct text-to-3D optimization toward a specific mode. Our ISD loss can be implemented by using IP-Adapter, a lightweight adapter for integrating image prompt capability to a text-to-image diffusion model, as a mode-selection module. A variant of this adapter, when not being prompted by a reference image, can serve as an efficient control variate to reduce variance in score estimates, thereby enhancing both output quality and optimization stability. Our experiments demonstrate that the ISD loss consistently achieves visually coherent, high-quality outputs and improves optimization speed compared to prior text-to-3D methods, as demonstrated through both qualitative and quantitative evaluations on the T3Bench benchmark suite.
comment: Project page: https://modedreamer.github.io/
♻ ☆ FoodMLLM-JP: Leveraging Multimodal Large Language Models for Japanese Recipe Generation
Research on food image understanding using recipe data has been a long-standing focus due to the diversity and complexity of the data. Moreover, food is inextricably linked to people's lives, making it a vital research area for practical applications such as dietary management. Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities, not only in their vast knowledge but also in their ability to handle languages naturally. While English is predominantly used, they can also support multiple languages including Japanese. This suggests that MLLMs are expected to significantly improve performance in food image understanding tasks. We fine-tuned open MLLMs LLaVA-1.5 and Phi-3 Vision on a Japanese recipe dataset and benchmarked their performance against the closed model GPT-4o. We then evaluated the content of generated recipes, including ingredients and cooking procedures, using 5,000 evaluation samples that comprehensively cover Japanese food culture. Our evaluation demonstrates that the open models trained on recipe data outperform GPT-4o, the current state-of-the-art model, in ingredient generation. Our model achieved F1 score of 0.531, surpassing GPT-4o's F1 score of 0.481, indicating a higher level of accuracy. Furthermore, our model exhibited comparable performance to GPT-4o in generating cooking procedure text.
comment: 15 pages, 5 figures. We found errors in the calculation of evaluation metrics, which were corrected in this version with $\color{blue}{\text{modifications highlighted in blue}}$. Please also see the Appendix
♻ ☆ Self-Supervised Iterative Refinement for Anomaly Detection in Industrial Quality Control
This study introduces the Iterative Refinement Process (IRP), a robust anomaly detection methodology designed for high-stakes industrial quality control. The IRP enhances defect detection accuracy through a cyclic data refinement strategy, iteratively removing misleading data points to improve model performance and robustness. We validate the IRP's effectiveness using two benchmark datasets, Kolektor SDD2 (KSDD2) and MVTec AD, covering a wide range of industrial products and defect types. Our experimental results demonstrate that the IRP consistently outperforms traditional anomaly detection models, particularly in environments with high noise levels. This study highlights the IRP's potential to significantly enhance anomaly detection processes in industrial settings, effectively managing the challenges of sparse and noisy data.
comment: Accepted to VISAPP 2025
♻ ☆ StarVid: Enhancing Semantic Alignment in Video Diffusion Models via Spatial and SynTactic Guided Attention Refocusing
Recent advances in text-to-video (T2V) generation with diffusion models have garnered significant attention. However, they typically perform well in scenes with a single object and motion, struggling in compositional scenarios with multiple objects and distinct motions to accurately reflect the semantic content of text prompts. To address these challenges, we propose \textbf{StarVid}, a plug-and-play, training-free method that improves semantic alignment between multiple subjects, their motions, and text prompts in T2V models. StarVid first leverages the spatial reasoning capabilities of large language models (LLMs) for two-stage motion trajectory planning based on text prompts. Such trajectories serve as spatial priors, guiding a spatial-aware loss to refocus cross-attention (CA) maps into distinctive regions. Furthermore, we propose a syntax-guided contrastive constraint to strengthen the correlation between the CA maps of verbs and their corresponding nouns, enhancing motion-subject binding. Both qualitative and quantitative evaluations demonstrate that the proposed framework significantly outperforms baseline methods, delivering videos of higher quality with improved semantic consistency.
♻ ☆ Towards Physically Realizable Adversarial Attacks in Embodied Vision Navigation IROS
The significant advancements in embodied vision navigation have raised concerns about its susceptibility to adversarial attacks exploiting deep neural networks. Investigating the adversarial robustness of embodied vision navigation is crucial, especially given the threat of 3D physical attacks that could pose risks to human safety. However, existing attack methods for embodied vision navigation often lack physical feasibility due to challenges in transferring digital perturbations into the physical world. Moreover, current physical attacks for object detection struggle to achieve both multi-view effectiveness and visual naturalness in navigation scenarios. To address this, we propose a practical attack method for embodied navigation by attaching adversarial patches to objects, where both opacity and textures are learnable. Specifically, to ensure effectiveness across varying viewpoints, we employ a multi-view optimization strategy based on object-aware sampling, which optimizes the patch's texture based on feedback from the vision-based perception model used in navigation. To make the patch inconspicuous to human observers, we introduce a two-stage opacity optimization mechanism, in which opacity is fine-tuned after texture optimization. Experimental results demonstrate that our adversarial patches decrease the navigation success rate by an average of 22.39%, outperforming previous methods in practicality, effectiveness, and naturalness. Code is available at: https://github.com/chen37058/Physical-Attacks-in-Embodied-Nav
comment: 7 pages, 7 figures, submitted to IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2025
♻ ☆ Efficient Learning With Sine-Activated Low-rank Matrices ICLR 2025
Low-rank decomposition has emerged as a vital tool for enhancing parameter efficiency in neural network architectures, gaining traction across diverse applications in machine learning. These techniques significantly lower the number of parameters, striking a balance between compactness and performance. However, a common challenge has been the compromise between parameter efficiency and the accuracy of the model, where reduced parameters often lead to diminished accuracy compared to their full-rank counterparts. In this work, we propose a novel theoretical framework that integrates a sinusoidal function within the low-rank decomposition process. This approach not only preserves the benefits of the parameter efficiency characteristic of low-rank methods but also increases the decomposition's rank, thereby enhancing model performance. Our method proves to be a plug in enhancement for existing low-rank models, as evidenced by its successful application in Vision Transformers (ViT), Large Language Models (LLMs), Neural Radiance Fields (NeRF) and 3D shape modelling.
comment: The first two authors contributed equally. Paper accepted at ICLR 2025
♻ ☆ A Decade's Battle on Dataset Bias: Are We There Yet? ICLR 2025
We revisit the "dataset classification" experiment suggested by Torralba & Efros (2011) a decade ago, in the new era with large-scale, diverse, and hopefully less biased datasets as well as more capable neural network architectures. Surprisingly, we observe that modern neural networks can achieve excellent accuracy in classifying which dataset an image is from: e.g., we report 84.7% accuracy on held-out validation data for the three-way classification problem consisting of the YFCC, CC, and DataComp datasets. Our further experiments show that such a dataset classifier could learn semantic features that are generalizable and transferable, which cannot be explained by memorization. We hope our discovery will inspire the community to rethink issues involving dataset bias.
comment: Published in ICLR 2025 (Oral Presentation)
♻ ☆ Saliency-Bench: A Comprehensive Benchmark for Evaluating Visual Explanations
Explainable AI (XAI) has gained significant attention for providing insights into the decision-making processes of deep learning models, particularly for image classification tasks through visual explanations visualized by saliency maps. Despite their success, challenges remain due to the lack of annotated datasets and standardized evaluation pipelines. In this paper, we introduce Saliency-Bench, a novel benchmark suite designed to evaluate visual explanations generated by saliency methods across multiple datasets. We curated, constructed, and annotated eight datasets, each covering diverse tasks such as scene classification, cancer diagnosis, object classification, and action classification, with corresponding ground-truth explanations. The benchmark includes a standardized and unified evaluation pipeline for assessing faithfulness and alignment of the visual explanation, providing a holistic visual explanation performance assessment. We benchmark these eight datasets with widely used saliency methods on different image classifier architectures to evaluate explanation quality. Additionally, we developed an easy-to-use API for automating the evaluation pipeline, from data accessing, and data loading, to result evaluation. The benchmark is available via our website: https://xaidataset.github.io.
Information Retrieval 19
Survey Perspective: The Role of Explainable AI in Threat Intelligence SIGIR
The increasing reliance on AI-based security tools in Security Operations Centers (SOCs) has transformed threat detection and response, yet analysts frequently struggle with alert overload, false positives, and lack of contextual relevance. The inability to effectively analyze AI-generated security alerts lead to inefficiencies in incident response and reduces trust in automated decision-making. In this paper, we show results and analysis of our investigation of how SOC analysts navigate AI-based alerts, their challenges with current security tools, and how explainability (XAI) integrated into their security workflows has the potential to become an effective decision support. In this vein, we conducted an industry survey. Using the survey responses, we analyze how security analysts' process, retrieve, and prioritize alerts. Our findings indicate that most analysts have not yet adopted XAI-integrated tools, but they express high interest in attack attribution, confidence scores, and feature contribution explanations to improve interpretability, and triage efficiency. Based on our findings, we also propose practical design recommendations for XAI-enhanced security alert systems, enabling AI-based cybersecurity solutions to be more transparent, interpretable, and actionable.
comment: 5 pages, SIGIR Symposium on IR in Practice (SIRIP), 2025
LLMInit: A Free Lunch from Large Language Models for Selective Initialization of Recommendation
Collaborative filtering models, particularly graph-based approaches, have demonstrated strong performance in capturing user-item interactions for recommendation systems. However, they continue to struggle in cold-start and data-sparse scenarios. The emergence of large language models (LLMs) like GPT and LLaMA presents new possibilities for enhancing recommendation performance, especially in cold-start settings. Despite their promise, LLMs pose challenges related to scalability and efficiency due to their high computational demands and limited ability to model complex user-item relationships effectively. In this work, we introduce a novel perspective on leveraging LLMs for CF model initialization. Through experiments, we uncover an embedding collapse issue when scaling CF models to larger embedding dimensions. To effectively harness large-scale LLM embeddings, we propose innovative selective initialization strategies utilizing random, uniform, and variance-based index sampling. Our comprehensive evaluation on multiple real-world datasets demonstrates significant performance gains across various CF models while maintaining a lower computational cost compared to existing LLM-based recommendation approaches.
☆ Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation
Many large-scale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show that sparse coding offers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We propose Contrastive Sparse Representation (CSR), a method that sparsifies pre-trained embeddings into a high-dimensional but selectively activated feature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed-often by large margins-while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount. Code is available at https://github.com/neilwen987/CSR_Adaptive_Rep
comment: A novel sparse coding framework designed for learning adaptive representation
☆ Retrieval Models Aren't Tool-Savvy: Benchmarking Tool Retrieval for Large Language Models
Tool learning aims to augment large language models (LLMs) with diverse tools, enabling them to act as agents for solving practical tasks. Due to the limited context length of tool-using LLMs, adopting information retrieval (IR) models to select useful tools from large toolsets is a critical initial step. However, the performance of IR models in tool retrieval tasks remains underexplored and unclear. Most tool-use benchmarks simplify this step by manually pre-annotating a small set of relevant tools for each task, which is far from the real-world scenarios. In this paper, we propose ToolRet, a heterogeneous tool retrieval benchmark comprising 7.6k diverse retrieval tasks, and a corpus of 43k tools, collected from existing datasets. We benchmark six types of models on ToolRet. Surprisingly, even the models with strong performance in conventional IR benchmarks, exhibit poor performance on ToolRet. This low retrieval quality degrades the task pass rate of tool-use LLMs. As a further step, we contribute a large-scale training dataset with over 200k instances, which substantially optimizes the tool retrieval ability of IR models.
☆ SAGE: A Framework of Precise Retrieval for RAG
Retrieval-augmented generation (RAG) has demonstrated significant proficiency in conducting question-answering (QA) tasks within a specified corpus. Nonetheless, numerous failure instances of RAG in QA still exist. These failures are not solely attributable to the limitations of Large Language Models (LLMs); instead, they predominantly arise from the retrieval of inaccurate information for LLMs due to two limitations: (1) Current RAG methods segment the corpus without considering semantics, making it difficult to find relevant context due to impaired correlation between questions and the segments. (2) There is a trade-off between missing essential context with fewer context retrieved and getting irrelevant context with more context retrieved. In this paper, we introduce a RAG framework (SAGE), to overcome these limitations. First, to address the segmentation issue without considering semantics, we propose to train a semantic segmentation model. This model is trained to segment the corpus into semantically complete chunks. Second, to ensure that only the most relevant chunks are retrieved while the irrelevant ones are ignored, we design a chunk selection algorithm to dynamically select chunks based on the decreasing speed of the relevance score, leading to a more relevant selection. Third, to further ensure the precision of the retrieved chunks, we propose letting LLMs assess whether retrieved chunks are excessive or lacking and then adjust the amount of context accordingly. Experiments show that SAGE outperforms baselines by 61.25% in the quality of QA on average. Moreover, by avoiding retrieving noisy context, SAGE lowers the cost of the tokens consumed in LLM inference and achieves a 49.41% enhancement in cost efficiency on average. Additionally, our work offers valuable insights for boosting RAG.
☆ Evaluating LLMs' Assessment of Mixed-Context Hallucination Through the Lens of Summarization
With the rapid development of large language models (LLMs), LLM-as-a-judge has emerged as a widely adopted approach for text quality evaluation, including hallucination evaluation. While previous studies have focused exclusively on single-context evaluation (e.g., discourse faithfulness or world factuality), real-world hallucinations typically involve mixed contexts, which remains inadequately evaluated. In this study, we use summarization as a representative task to comprehensively evaluate LLMs' capability in detecting mixed-context hallucinations, specifically distinguishing between factual and non-factual hallucinations. Through extensive experiments across direct generation and retrieval-based models of varying scales, our main observations are: (1) LLMs' intrinsic knowledge introduces inherent biases in hallucination evaluation; (2) These biases particularly impact the detection of factual hallucinations, yielding a significant performance bottleneck; (3) The fundamental challenge lies in effective knowledge utilization, balancing between LLMs' intrinsic knowledge and external context for accurate mixed-context hallucination evaluation.
comment: 8 pages, 5 figures for main body
☆ CoPL: Collaborative Preference Learning for Personalizing LLMs
Personalizing large language models (LLMs) is important for aligning outputs with diverse user preferences, yet existing methods struggle with flexibility and generalization. We propose CoPL (Collaborative Preference Learning), a graph-based collaborative filtering framework that models user-response relationships to enhance preference estimation, particularly in sparse annotation settings. By integrating a mixture of LoRA experts, CoPL efficiently fine-tunes LLMs while dynamically balancing shared and user-specific preferences. Additionally, an optimization-free adaptation strategy enables generalization to unseen users without fine-tuning. Experiments on UltraFeedback-P demonstrate that CoPL outperforms existing personalized reward models, effectively capturing both common and controversial preferences, making it a scalable solution for personalized LLM alignment.
comment: 13pages, 4 figures, 6tables
☆ Leveraging LLMs for Mental Health: Detection and Recommendations from Social Discussions
Textual data from social platforms captures various aspects of mental health through discussions around and across issues, while users reach out for help and others sympathize and offer support. We propose a comprehensive framework that leverages Natural Language Processing (NLP) and Generative AI techniques to identify and assess mental health disorders, detect their severity, and create recommendations for behavior change and therapeutic interventions based on users' posts on Reddit. To classify the disorders, we use rule-based labeling methods as well as advanced pre-trained NLP models to extract nuanced semantic features from the data. We fine-tune domain-adapted and generic pre-trained NLP models based on predictions from specialized Large Language Models (LLMs) to improve classification accuracy. Our hybrid approach combines the generalization capabilities of pre-trained models with the domain-specific insights captured by LLMs, providing an improved understanding of mental health discourse. Our findings highlight the strengths and limitations of each model, offering valuable insights into their practical applicability. This research potentially facilitates early detection and personalized care to aid practitioners and aims to facilitate timely interventions and improve overall well-being, thereby contributing to the broader field of mental health surveillance and digital health analytics.
comment: 5 pages, 4 figures, 3 tables, to be published in WI-IAT 2024
☆ Streaming Piano Transcription Based on Consistent Onset and Offset Decoding with Sustain Pedal Detection
This paper describes a streaming audio-to-MIDI piano transcription approach that aims to sequentially translate a music signal into a sequence of note onset and offset events. The sequence-to-sequence nature of this task may call for the computationally-intensive transformer model for better performance, which has recently been used for offline transcription benchmarks and could be extended for streaming transcription with causal attention mechanisms. We assume that the performance limitation of this naive approach lies in the decoder. Although time-frequency features useful for onset detection are considerably different from those for offset detection, the single decoder is trained to output a mixed sequence of onset and offset events without guarantee of the correspondence between the onset and offset events of the same note. To overcome this limitation, we propose a streaming encoder-decoder model that uses a convolutional encoder aggregating local acoustic features, followed by an autoregressive Transformer decoder detecting a variable number of onset events and another decoder detecting the offset events for the active pitches with validation of the sustain pedal at each time frame. Experiments using the MAESTRO dataset showed that the proposed streaming method performed comparably with or even better than the state-of-the-art offline methods while significantly reducing the computational cost.
comment: Accepted to ISMIR 2024
☆ SRAG: Structured Retrieval-Augmented Generation for Multi-Entity Question Answering over Wikipedia Graph
Multi-entity question answering (MEQA) poses significant challenges for large language models (LLMs), which often struggle to consolidate scattered information across multiple documents. An example question might be "What is the distribution of IEEE Fellows among various fields of study?", which requires retrieving information from diverse sources e.g., Wikipedia pages. The effectiveness of current retrieval-augmented generation (RAG) methods is limited by the LLMs' capacity to aggregate insights from numerous pages. To address this gap, this paper introduces a structured RAG (SRAG) framework that systematically organizes extracted entities into relational tables (e.g., tabulating entities with schema columns like "name" and "field of study") and then apply table-based reasoning techniques. Our approach decouples retrieval and reasoning, enabling LLMs to focus on structured data analysis rather than raw text aggregation. Extensive experiments on Wikipedia-based multi-entity QA tasks demonstrate that SRAG significantly outperforms state-of-the-art long-context LLMs and RAG solutions, achieving a 29.6% improvement in accuracy. The results underscore the efficacy of structuring unstructured data to enhance LLMs' reasoning capabilities.
☆ Composed Multi-modal Retrieval: A Survey of Approaches and Applications
With the rapid growth of multi-modal data from social media, short video platforms, and e-commerce, content-based retrieval has become essential for efficiently searching and utilizing heterogeneous information. Over time, retrieval techniques have evolved from Unimodal Retrieval (UR) to Cross-modal Retrieval (CR) and, more recently, to Composed Multi-modal Retrieval (CMR). CMR enables users to retrieve images or videos by integrating a reference visual input with textual modifications, enhancing search flexibility and precision. This paper provides a comprehensive review of CMR, covering its fundamental challenges, technical advancements, and categorization into supervised, zero-shot, and semi-supervised learning paradigms. We discuss key research directions, including data augmentation, model architecture, and loss optimization in supervised CMR, as well as transformation frameworks and external knowledge integration in zero-shot CMR. Additionally, we highlight the application potential of CMR in composed image retrieval, video retrieval, and person retrieval, which have significant implications for e-commerce, online search, and public security. Given its ability to refine and personalize search experiences, CMR is poised to become a pivotal technology in next-generation retrieval systems. A curated list of related works and resources is available at: https://github.com/kkzhang95/Awesome-Composed-Multi-modal-Retrieval
☆ HI-Series Algorithms A Hybrid of Substance Diffusion Algorithm and Collaborative Filtering
Recommendation systems face the challenge of balancing accuracy and diversity, as traditional collaborative filtering (CF) and network-based diffusion algorithms exhibit complementary limitations. While item-based CF (ItemCF) enhances diversity through item similarity, it compromises accuracy. Conversely, mass diffusion (MD) algorithms prioritize accuracy by favoring popular items but lack diversity. To address this trade-off, we propose the HI-series algorithms, hybrid models integrating ItemCF with diffusion-based approaches (MD, HHP, BHC, BD) through a nonlinear combination controlled by parameter $\epsilon$. This hybridization leverages ItemCF's diversity and MD's accuracy, extending to advanced diffusion models (HI-HHP, HI-BHC, HI-BD) for enhanced performance. Experiments on MovieLens, Netflix, and RYM datasets demonstrate that HI-series algorithms significantly outperform their base counterparts. In sparse data ($20\%$ training), HI-MD achieves a $0.8\%$-$4.4\%$ improvement in F1-score over MD while maintaining higher diversity (Diversity@20: 459 vs. 396 on MovieLens). For dense data ($80\%$ training), HI-BD improves F1-score by $2.3\%$-$5.2\%$ compared to BD, with diversity gains up to $18.6\%$. Notably, hybrid models consistently enhance novelty in sparse settings and exhibit robust parameter adaptability. The results validate that strategic hybridization effectively breaks the accuracy-diversity trade-off, offering a flexible framework for optimizing recommendation systems across data sparsity levels.
☆ Cancer Type, Stage and Prognosis Assessment from Pathology Reports using LLMs
Large Language Models (LLMs) have shown significant promise across various natural language processing tasks. However, their application in the field of pathology, particularly for extracting meaningful insights from unstructured medical texts such as pathology reports, remains underexplored and not well quantified. In this project, we leverage state-of-the-art language models, including the GPT family, Mistral models, and the open-source Llama models, to evaluate their performance in comprehensively analyzing pathology reports. Specifically, we assess their performance in cancer type identification, AJCC stage determination, and prognosis assessment, encompassing both information extraction and higher-order reasoning tasks. Based on a detailed analysis of their performance metrics in a zero-shot setting, we developed two instruction-tuned models: Path-llama3.1-8B and Path-GPT-4o-mini-FT. These models demonstrated superior performance in zero-shot cancer type identification, staging, and prognosis assessment compared to the other models evaluated.
☆ ReaderLM-v2: Small Language Model for HTML to Markdown and JSON
We present ReaderLM-v2, a compact 1.5 billion parameter language model designed for efficient web content extraction. Our model processes documents up to 512K tokens, transforming messy HTML into clean Markdown or JSON formats with high accuracy -- making it an ideal tool for grounding large language models. The model's effectiveness results from two key innovations: (1) a three-stage data synthesis pipeline that generates high quality, diverse training data by iteratively drafting, refining, and critiquing web content extraction; and (2) a unified training framework combining continuous pre-training with multi-objective optimization. Intensive evaluation demonstrates that ReaderLM-v2 outperforms GPT-4o-2024-08-06 and other larger models by 15-20\% on carefully curated benchmarks, particularly excelling at documents exceeding 100K tokens, while maintaining significantly lower computational requirements.
comment: 9 pages, 10-12 refs
♻ ☆ Evaluating Intelligence via Trial and Error
Intelligence is a crucial trait for species to find solutions within a limited number of trial-and-error attempts. Building on this idea, we introduce Survival Game as a framework to evaluate intelligence based on the number of failed attempts in a trial-and-error process. Fewer failures indicate higher intelligence. When the expectation and variance of failure counts are both finite, it signals the ability to consistently find solutions to new challenges, which we define as the Autonomous Level of intelligence. Using Survival Game, we comprehensively evaluate existing AI systems. Our results show that while AI systems achieve the Autonomous Level in simple tasks, they are still far from it in more complex tasks, such as vision, search, recommendation, and language. While scaling current AI technologies might help, this would come at an astronomical cost. Projections suggest that achieving the Autonomous Level for general tasks would require $10^{26}$ parameters. To put this into perspective, loading such a massive model requires so many H100 GPUs that their total value is $10^{7}$ times that of Apple Inc.'s market value. Even with Moore's Law, supporting such a parameter scale would take $70$ years. This staggering cost highlights the complexity of human tasks and the inadequacies of current AI technologies. To further investigate this phenomenon, we conduct a theoretical analysis of Survival Game and its experimental results. Our findings suggest that human tasks possess a criticality property. As a result, Autonomous Level requires a deep understanding of the task's underlying mechanisms. Current AI systems, however, do not fully grasp these mechanisms and instead rely on superficial mimicry, making it difficult for them to reach an autonomous level. We believe Survival Game can not only guide the future development of AI but also offer profound insights into human intelligence.
♻ ☆ Optimize Incompatible Parameters through Compatibility-aware Knowledge Integration AAAI'25
Deep neural networks have become foundational to advancements in multiple domains, including recommendation systems, natural language processing, and so on. Despite their successes, these models often contain incompatible parameters that can be underutilized or detrimental to model performance, particularly when faced with specific, varying data distributions. Existing research excels in removing such parameters or merging the outputs of multiple different pretrained models. However, the former focuses on efficiency rather than performance, while the latter requires several times more computing and storage resources to support inference. In this paper, we set the goal to explicitly improve these incompatible parameters by leveraging the complementary strengths of different models, thereby directly enhancing the models without any additional parameters. Specifically, we propose Compatibility-aware Knowledge Integration (CKI), which consists of Parameter Compatibility Assessment and Parameter Splicing, which are used to evaluate the knowledge content of multiple models and integrate the knowledge into one model, respectively. The integrated model can be used directly for inference or for further fine-tuning. We conduct extensive experiments on various datasets for recommendation and language tasks, and the results show that Compatibility-aware Knowledge Integration can effectively optimize incompatible parameters under multiple tasks and settings to break through the training limit of the original model without increasing the inference cost.
comment: Published on AAAI'25(Oral): The Annual AAAI Conference on Artificial Intelligence
♻ ☆ External Large Foundation Model: How to Efficiently Serve Trillions of Parameters for Online Ads Recommendation WWW
Ads recommendation is a prominent service of online advertising systems and has been actively studied. Recent studies indicate that scaling-up and advanced design of the recommendation model can bring significant performance improvement. However, with a larger model scale, such prior studies have a significantly increasing gap from industry as they often neglect two fundamental challenges in industrial-scale applications. First, training and inference budgets are restricted for the model to be served, exceeding which may incur latency and impair user experience. Second, large-volume data arrive in a streaming mode with data distributions dynamically shifting, as new users/ads join and existing users/ads leave the system. We propose the External Large Foundation Model (ExFM) framework to address the overlooked challenges. Specifically, we develop external distillation and a data augmentation system (DAS) to control the computational cost of training/inference while maintaining high performance. We design the teacher in a way like a foundation model (FM) that can serve multiple students as vertical models (VMs) to amortize its building cost. We propose Auxiliary Head and Student Adapter to mitigate the data distribution gap between FM and VMs caused by the streaming data issue. Comprehensive experiments on internal industrial-scale applications and public datasets demonstrate significant performance gain by ExFM.
comment: Accepted by the ACM Web Conference (WWW) 2025 Industrial Track as Oral Presentation
♻ ☆ RAGEval: Scenario Specific RAG Evaluation Dataset Generation Framework
Retrieval-Augmented Generation (RAG) is a powerful approach that enables large language models (LLMs) to incorporate external knowledge. However, evaluating the effectiveness of RAG systems in specialized scenarios remains challenging due to the high costs of data construction and the lack of suitable evaluation metrics. This paper introduces RAGEval, a framework designed to assess RAG systems across diverse scenarios by generating high-quality documents, questions, answers, and references through a schema-based pipeline. With a focus on factual accuracy, we propose three novel metrics: Completeness, Hallucination, and Irrelevance to evaluate LLM generated responses rigorously. Experimental results show that RAGEval outperforms zero-shot and one-shot methods in terms of clarity, safety, conformity, and richness of generated samples. Furthermore, the use of LLMs for scoring the proposed metrics demonstrates a high level of consistency with human evaluations. RAGEval establishes a new paradigm for evaluating RAG systems in real-world applications. The code and dataset are released at https://github.com/OpenBMB/RAGEval.
comment: https://github.com/OpenBMB/RAGEval
♻ ☆ TEARS: Textual Representations for Scrutable Recommendations
Traditional recommender systems rely on high-dimensional (latent) embeddings for modeling user-item interactions, often resulting in opaque representations that lack interpretability. Moreover, these systems offer limited control to users over their recommendations. Inspired by recent work, we introduce TExtuAl Representations for Scrutable recommendations (TEARS) to address these challenges. Instead of representing a user's interests through a latent embedding, TEARS encodes them in natural text, providing transparency and allowing users to edit them. To do so, TEARS uses a modern LLM to generate user summaries based on user preferences. We find the summaries capture user preferences uniquely. Using these summaries, we take a hybrid approach where we use an optimal transport procedure to align the summaries' representation with the learned representation of a standard VAE for collaborative filtering. We find this approach can surpass the performance of three popular VAE models while providing user-controllable recommendations. We also analyze the controllability of TEARS through three simulated user tasks to evaluate the effectiveness of a user editing its summary.
Machine Learning 105
♻ ☆ Langevin Multiplicative Weights Update with Applications in Polynomial Portfolio Management AAAI-2025
We consider nonconvex optimization problem over simplex, and more generally, a product of simplices. We provide an algorithm, Langevin Multiplicative Weights Update (LMWU) for solving global optimization problems by adding a noise scaling with the non-Euclidean geometry in the simplex. Non-convex optimization has been extensively studied by machine learning community due to its application in various scenarios such as neural network approximation and finding Nash equilibrium. Despite recent progresses on provable guarantee of escaping and avoiding saddle point (convergence to local minima) and global convergence of Langevin gradient based method without constraints, the global optimization with constraints is less studied. We show that LMWU algorithm is provably convergent to interior global minima with a non-asymptotic convergence analysis. We verify the efficiency of the proposed algorithm in real data set from polynomial portfolio management, where optimization of a highly non-linear objective function plays a crucial role.
comment: Accepted for AAAI-2025
♻ ☆ Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?
The advent of test-time scaling in large language models (LLMs), exemplified by OpenAI's o1 series, has advanced reasoning capabilities by scaling computational resource allocation during inference. While successors like QwQ, Deepseek-R1 (R1) and LIMO replicate these advancements, whether these models truly possess test-time scaling capabilities remains underexplored. This study found that longer CoTs of these o1-like models do not consistently enhance accuracy; in fact, correct solutions are often shorter than incorrect ones for the same questions. Further investigation shows this phenomenon is closely related to models' self-revision capabilities - longer CoTs contain more self-revisions, which often lead to performance degradation. We then compare sequential and parallel scaling strategies on QwQ, R1 and LIMO, finding that parallel scaling achieves better coverage and scalability. Based on these insights, we propose Shortest Majority Vote, a method that combines parallel scaling strategies with CoT length characteristics, significantly improving models' test-time scalability compared to conventional majority voting approaches.
comment: Add the github link
♻ ☆ Gradient-Based Multi-Objective Deep Learning: Algorithms, Theories, Applications, and Beyond
Multi-objective optimization (MOO) in deep learning aims to simultaneously optimize multiple conflicting objectives, a challenge frequently encountered in areas like multi-task learning and multi-criteria learning. Recent advancements in gradient-based MOO methods have enabled the discovery of diverse types of solutions, ranging from a single balanced solution to finite or even infinite Pareto sets, tailored to user needs. These developments have broad applications across domains such as reinforcement learning, computer vision, recommendation systems, and large language models. This survey provides the first comprehensive review of gradient-based MOO in deep learning, covering algorithms, theories, and practical applications. By unifying various approaches and identifying critical challenges, it serves as a foundational resource for driving innovation in this evolving field. A comprehensive list of MOO algorithms in deep learning is available at https://github.com/Baijiong-Lin/Awesome-Multi-Objective-Deep-Learning.
♻ ☆ Preconditioned Inexact Stochastic ADMM for Deep Model
The recent advancement of foundation models (FMs) has brought about a paradigm shift, revolutionizing various sectors worldwide. The popular optimizers used to train these models are stochastic gradient descent-based algorithms, which face inherent limitations, such as slow convergence and stringent assumptions for convergence. In particular, data heterogeneity arising from distributed settings poses significant challenges to their theoretical and numerical performance. This paper develops an algorithm, PISA ({P}reconditioned {I}nexact {S}tochastic {A}lternating Direction Method of Multipliers), which enables scalable parallel computing and supports various second-moment schemes. Grounded in rigorous theoretical guarantees, the algorithm converges under the sole assumption of Lipschitz continuity of the gradient, thereby removing the need for other conditions commonly imposed by stochastic methods. This capability enables PISA to tackle the challenge of data heterogeneity effectively. Comprehensive experimental evaluations for training or fine-tuning diverse FMs, including vision models, large language models, reinforcement learning models, generative adversarial networks, and recurrent neural networks, demonstrate its superior numerical performance compared to various state-of-the-art optimizers.
♻ ☆ Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks ICLR 2025
While large models trained with self-supervised learning on offline datasets have shown remarkable capabilities in text and image domains, achieving the same generalisation for agents that act in sequential decision problems remains an open challenge. In this work, we take a step towards this goal by procedurally generating tens of millions of 2D physics-based tasks and using these to train a general reinforcement learning (RL) agent for physical control. To this end, we introduce Kinetix: an open-ended space of physics-based RL environments that can represent tasks ranging from robotic locomotion and grasping to video games and classic RL environments, all within a unified framework. Kinetix makes use of our novel hardware-accelerated physics engine Jax2D that allows us to cheaply simulate billions of environment steps during training. Our trained agent exhibits strong physical reasoning capabilities in 2D space, being able to zero-shot solve unseen human-designed environments. Furthermore, fine-tuning this general agent on tasks of interest shows significantly stronger performance than training an RL agent *tabula rasa*. This includes solving some environments that standard RL training completely fails at. We believe this demonstrates the feasibility of large scale, mixed-quality pre-training for online RL and we hope that Kinetix will serve as a useful framework to investigate this further.
comment: ICLR 2025 Oral. The first two authors contributed equally. Project page located at: https://kinetix-env.github.io/
♻ ☆ Optimizing Backward Policies in GFlowNets via Trajectory Likelihood Maximization ICLR 2025
Generative Flow Networks (GFlowNets) are a family of generative models that learn to sample objects with probabilities proportional to a given reward function. The key concept behind GFlowNets is the use of two stochastic policies: a forward policy, which incrementally constructs compositional objects, and a backward policy, which sequentially deconstructs them. Recent results show a close relationship between GFlowNet training and entropy-regularized reinforcement learning (RL) problems with a particular reward design. However, this connection applies only in the setting of a fixed backward policy, which might be a significant limitation. As a remedy to this problem, we introduce a simple backward policy optimization algorithm that involves direct maximization of the value function in an entropy-regularized Markov Decision Process (MDP) over intermediate rewards. We provide an extensive experimental evaluation of the proposed approach across various benchmarks in combination with both RL and GFlowNet algorithms and demonstrate its faster convergence and mode discovery in complex environments.
comment: ICLR 2025
♻ ☆ Optimize Incompatible Parameters through Compatibility-aware Knowledge Integration AAAI'25
Deep neural networks have become foundational to advancements in multiple domains, including recommendation systems, natural language processing, and so on. Despite their successes, these models often contain incompatible parameters that can be underutilized or detrimental to model performance, particularly when faced with specific, varying data distributions. Existing research excels in removing such parameters or merging the outputs of multiple different pretrained models. However, the former focuses on efficiency rather than performance, while the latter requires several times more computing and storage resources to support inference. In this paper, we set the goal to explicitly improve these incompatible parameters by leveraging the complementary strengths of different models, thereby directly enhancing the models without any additional parameters. Specifically, we propose Compatibility-aware Knowledge Integration (CKI), which consists of Parameter Compatibility Assessment and Parameter Splicing, which are used to evaluate the knowledge content of multiple models and integrate the knowledge into one model, respectively. The integrated model can be used directly for inference or for further fine-tuning. We conduct extensive experiments on various datasets for recommendation and language tasks, and the results show that Compatibility-aware Knowledge Integration can effectively optimize incompatible parameters under multiple tasks and settings to break through the training limit of the original model without increasing the inference cost.
comment: Published on AAAI'25(Oral): The Annual AAAI Conference on Artificial Intelligence
♻ ☆ AnyECG: Foundational Models for Multitask Cardiac Analysis in Real-World Settings
Electrocardiogram (ECG), a non-invasive and affordable tool for cardiac monitoring, is highly sensitive in detecting acute heart attacks. However, due to the lengthy nature of ECG recordings, numerous machine learning methods have been developed for automated heart disease detection to reduce human workload. Despite these efforts, performance remains suboptimal. A key obstacle is the inherent complexity of ECG data, which includes heterogeneity (e.g., varying sampling rates), high levels of noise, demographic-related pattern shifts, and intricate rhythm-event associations. To overcome these challenges, this paper introduces AnyECG, a foundational model designed to extract robust representations from any real-world ECG data. Specifically, a tailored ECG Tokenizer encodes each fixed-duration ECG fragment into a token and, guided by proxy tasks, converts noisy, continuous ECG features into discrete, compact, and clinically meaningful local rhythm codes. These codes encapsulate basic morphological, frequency, and demographic information (e.g., sex), effectively mitigating signal noise. We further pre-train the AnyECG to learn rhythmic pattern associations across ECG tokens, enabling the capture of cardiac event semantics. By being jointly pre-trained on diverse ECG data sources, AnyECG is capable of generalizing across a wide range of downstream tasks where ECG signals are recorded from various devices and scenarios. The experimental results show that AnyECG achieves an average performance improvement of 6% across four critical tasks-anomaly detection, arrhythmia classification, corrupted lead generation, and ultra-long ECG recognition. AnyECG learns common ECG rhythm from data and significantly outperforms state-of-the-art methods in each of these tasks.
♻ ☆ Nonasymptotic Analysis of Stochastic Gradient Descent with the Richardson-Romberg Extrapolation ICLR-2025
We address the problem of solving strongly convex and smooth minimization problems using stochastic gradient descent (SGD) algorithm with a constant step size. Previous works suggested to combine the Polyak-Ruppert averaging procedure with the Richardson-Romberg extrapolation to reduce the asymptotic bias of SGD at the expense of a mild increase of the variance. We significantly extend previous results by providing an expansion of the mean-squared error of the resulting estimator with respect to the number of iterations $n$. We show that the root mean-squared error can be decomposed into the sum of two terms: a leading one of order $\mathcal{O}(n^{-1/2})$ with explicit dependence on a minimax-optimal asymptotic covariance matrix, and a second-order term of order $\mathcal{O}(n^{-3/4})$, where the power $3/4$ is best known. We also extend this result to the higher-order moment bounds. Our analysis relies on the properties of the SGD iterates viewed as a time-homogeneous Markov chain. In particular, we establish that this chain is geometrically ergodic with respect to a suitably defined weighted Wasserstein semimetric.
comment: ICLR-2025, camera-ready version
♻ ☆ MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypotheses ICLR 2025
Scientific discovery contributes largely to human society's prosperity, and recent progress shows that LLMs could potentially catalyze this process. However, it is still unclear whether LLMs can discover novel and valid hypotheses in chemistry. In this work, we investigate this central research question: Can LLMs automatically discover novel and valid chemistry research hypotheses given only a chemistry research background (consisting of a research question and/or a background survey), without limitation on the domain of the research question? After extensive discussions with chemistry experts, we propose an assumption that a majority of chemistry hypotheses can be resulted from a research background and several inspirations. With this key insight, we break the central question into three smaller fundamental questions. In brief, they are: (1) given a background question, whether LLMs can retrieve good inspirations; (2) with background and inspirations, whether LLMs can lead to hypothesis; and (3) whether LLMs can identify good hypotheses to rank them higher. To investigate these questions, we construct a benchmark consisting of 51 chemistry papers published in Nature, Science, or a similar level in 2024 (all papers are only available online since 2024). Every paper is divided by chemistry PhD students into three components: background, inspirations, and hypothesis. The goal is to rediscover the hypothesis, given only the background and a large randomly selected chemistry literature corpus consisting the ground truth inspiration papers, with LLMs trained with data up to 2023. We also develop an LLM-based multi-agent framework that leverages the assumption, consisting of three stages reflecting the three smaller questions. The proposed method can rediscover many hypotheses with very high similarity with the ground truth ones, covering the main innovations.
comment: Accepted by ICLR 2025
♻ ☆ A Meta-Learning Approach to Bayesian Causal Discovery
Discovering a unique causal structure is difficult due to both inherent identifiability issues, and the consequences of finite data. As such, uncertainty over causal structures, such as those obtained from a Bayesian posterior, are often necessary for downstream tasks. Finding an accurate approximation to this posterior is challenging, due to the large number of possible causal graphs, as well as the difficulty in the subproblem of finding posteriors over the functional relationships of the causal edges. Recent works have used meta-learning to view the problem of estimating the maximum a-posteriori causal graph as supervised learning. Yet, these methods are limited when estimating the full posterior as they fail to encode key properties of the posterior, such as correlation between edges and permutation equivariance with respect to nodes. Further, these methods also cannot reliably sample from the posterior over causal structures. To address these limitations, we propose a Bayesian meta learning model that allows for sampling causal structures from the posterior and encodes these key properties. We compare our meta-Bayesian causal discovery against existing Bayesian causal discovery methods, demonstrating the advantages of directly learning a posterior over causal structure.
♻ ☆ Poison-splat: Computation Cost Attack on 3D Gaussian Splatting ICLR 2025
3D Gaussian splatting (3DGS), known for its groundbreaking performance and efficiency, has become a dominant 3D representation and brought progress to many 3D vision tasks. However, in this work, we reveal a significant security vulnerability that has been largely overlooked in 3DGS: the computation cost of training 3DGS could be maliciously tampered by poisoning the input data. By developing an attack named Poison-splat, we reveal a novel attack surface where the adversary can poison the input images to drastically increase the computation memory and time needed for 3DGS training, pushing the algorithm towards its worst computation complexity. In extreme cases, the attack can even consume all allocable memory, leading to a Denial-of-Service (DoS) that disrupts servers, resulting in practical damages to real-world 3DGS service vendors. Such a computation cost attack is achieved by addressing a bi-level optimization problem through three tailored strategies: attack objective approximation, proxy model rendering, and optional constrained optimization. These strategies not only ensure the effectiveness of our attack but also make it difficult to defend with simple defensive measures. We hope the revelation of this novel attack surface can spark attention to this crucial yet overlooked vulnerability of 3DGS systems. Our code is available at https://github.com/jiahaolu97/poison-splat .
comment: Accepted by ICLR 2025 as a spotlight paper
♻ ☆ On the Geometry and Optimization of Polynomial Convolutional Networks AISTATS 2025
We study convolutional neural networks with monomial activation functions. Specifically, we prove that their parameterization map is regular and is an isomorphism almost everywhere, up to rescaling the filters. By leveraging on tools from algebraic geometry, we explore the geometric properties of the image in function space of this map - typically referred to as neuromanifold. In particular, we compute the dimension and the degree of the neuromanifold, which measure the expressivity of the model, and describe its singularities. Moreover, for a generic large dataset, we derive an explicit formula that quantifies the number of critical points arising in the optimization of a regression loss.
comment: Accepted at AISTATS 2025
♻ ☆ Federated Temporal Graph Clustering
Temporal graph clustering is a complex task that involves discovering meaningful structures in dynamic graphs where relationships and entities change over time. Existing methods typically require centralized data collection, which poses significant privacy and communication challenges. In this work, we introduce a novel Federated Temporal Graph Clustering (FTGC) framework that enables decentralized training of graph neural networks (GNNs) across multiple clients, ensuring data privacy throughout the process. Our approach incorporates a temporal aggregation mechanism to effectively capture the evolution of graph structures over time and a federated optimization strategy to collaboratively learn high-quality clustering representations. By preserving data privacy and reducing communication overhead, our framework achieves competitive performance on temporal graph datasets, making it a promising solution for privacy-sensitive, real-world applications involving dynamic data.
comment: 8 pages, 1 figure
♻ ☆ MLOmics: Benchmark for Machine Learning on Cancer Multi-Omics Data
Framing the investigation of diverse cancers as a machine learning problem has recently shown significant potential in multi-omics analysis and cancer research. Empowering these successful machine learning models are the high-quality training datasets with sufficient data volume and adequate preprocessing. However, while there exist several public data portals including The Cancer Genome Atlas (TCGA) multi-omics initiative or open-bases such as the LinkedOmics, these databases are not off-the-shelf for existing machine learning models. In this paper we propose MLOmics, an open cancer multi-omics benchmark aiming at serving better the development and evaluation of bioinformatics and machine learning models. MLOmics contains 8,314 patient samples covering all 32 cancer types with four omics types, stratified features, and extensive baselines. Complementary support for downstream analysis and bio-knowledge linking are also included to support interdisciplinary analysis.
comment: Under review
♻ ☆ Optimal Brain Apoptosis ICLR 2025
The increasing complexity and parameter count of Convolutional Neural Networks (CNNs) and Transformers pose challenges in terms of computational efficiency and resource demands. Pruning has been identified as an effective strategy to address these challenges by removing redundant elements such as neurons, channels, or connections, thereby enhancing computational efficiency without heavily compromising performance. This paper builds on the foundational work of Optimal Brain Damage (OBD) by advancing the methodology of parameter importance estimation using the Hessian matrix. Unlike previous approaches that rely on approximations, we introduce Optimal Brain Apoptosis (OBA), a novel pruning method that calculates the Hessian-vector product value directly for each parameter. By decomposing the Hessian matrix across network layers and identifying conditions under which inter-layer Hessian submatrices are non-zero, we propose a highly efficient technique for computing the second-order Taylor expansion of parameters. This approach allows for a more precise pruning process, particularly in the context of CNNs and Transformers, as validated in our experiments including VGG19, ResNet32, ResNet50, and ViT-B/16 on CIFAR10, CIFAR100 and Imagenet datasets. Our code is available at https://github.com/NEU-REAL/OBA.
comment: Accepted to ICLR 2025
♻ ☆ Exploring the Effectiveness of Object-Centric Representations in Visual Question Answering: Comparative Insights with Foundation Models ICLR 2025
Object-centric (OC) representations, which model visual scenes as compositions of discrete objects, have the potential to be used in various downstream tasks to achieve systematic compositional generalization and facilitate reasoning. However, these claims have yet to be thoroughly validated empirically. Recently, foundation models have demonstrated unparalleled capabilities across diverse domains, from language to computer vision, positioning them as a potential cornerstone of future research for a wide range of computational tasks. In this paper, we conduct an extensive empirical study on representation learning for downstream Visual Question Answering (VQA), which requires an accurate compositional understanding of the scene. We thoroughly investigate the benefits and trade-offs of OC models and alternative approaches including large pre-trained foundation models on both synthetic and real-world data, ultimately identifying a promising path to leverage the strengths of both paradigms. The extensiveness of our study, encompassing over 600 downstream VQA models and 15 different types of upstream representations, also provides several additional insights that we believe will be of interest to the community at large.
comment: Published at ICLR 2025
♻ ☆ Adaptive $Q$-Network: On-the-fly Target Selection for Deep Reinforcement Learning ICLR
Deep Reinforcement Learning (RL) is well known for being highly sensitive to hyperparameters, requiring practitioners substantial efforts to optimize them for the problem at hand. This also limits the applicability of RL in real-world scenarios. In recent years, the field of automated Reinforcement Learning (AutoRL) has grown in popularity by trying to address this issue. However, these approaches typically hinge on additional samples to select well-performing hyperparameters, hindering sample-efficiency and practicality. Furthermore, most AutoRL methods are heavily based on already existing AutoML methods, which were originally developed neglecting the additional challenges inherent to RL due to its non-stationarities. In this work, we propose a new approach for AutoRL, called Adaptive $Q$-Network (AdaQN), that is tailored to RL to take into account the non-stationarity of the optimization procedure without requiring additional samples. AdaQN learns several $Q$-functions, each one trained with different hyperparameters, which are updated online using the $Q$-function with the smallest approximation error as a shared target. Our selection scheme simultaneously handles different hyperparameters while coping with the non-stationarity induced by the RL optimization procedure and being orthogonal to any critic-based RL algorithm. We demonstrate that AdaQN is theoretically sound and empirically validate it in MuJoCo control problems and Atari $2600$ games, showing benefits in sample-efficiency, overall performance, robustness to stochasticity and training stability.
comment: Accepted at ICLR https://iclr.cc/virtual/2025/poster/28508
♻ ☆ Offline Model-Based Optimization by Learning to Rank ICLR 2025
Offline model-based optimization (MBO) aims to identify a design that maximizes a black-box function using only a fixed, pre-collected dataset of designs and their corresponding scores. A common approach in offline MBO is to train a regression-based surrogate model by minimizing mean squared error (MSE) and then find the best design within this surrogate model by different optimizers (e.g., gradient ascent). However, a critical challenge is the risk of out-of-distribution errors, i.e., the surrogate model may typically overestimate the scores and mislead the optimizers into suboptimal regions. Prior works have attempted to address this issue in various ways, such as using regularization techniques and ensemble learning to enhance the robustness of the model, but it still remains. In this paper, we argue that regression models trained with MSE are not well-aligned with the primary goal of offline MBO, which is to select promising designs rather than to predict their scores precisely. Notably, if a surrogate model can maintain the order of candidate designs based on their relative score relationships, it can produce the best designs even without precise predictions. To validate it, we conduct experiments to compare the relationship between the quality of the final designs and MSE, finding that the correlation is really very weak. In contrast, a metric that measures order-maintaining quality shows a significantly stronger correlation. Based on this observation, we propose learning a ranking-based model that leverages learning to rank techniques to prioritize promising designs based on their relative scores. We show that the generalization error on ranking loss can be well bounded. Empirical results across diverse tasks demonstrate the superior performance of our proposed ranking-based models than twenty existing methods.
comment: ICLR 2025
♻ ☆ Variational Best-of-N Alignment
Best-of-N (BoN) is a popular and effective algorithm for aligning language models to human preferences. The algorithm works as follows: at inference time, N samples are drawn from the language model, and the sample with the highest reward, as judged by a reward model, is returned as the output. Despite its effectiveness, BoN is computationally expensive; it reduces sampling throughput by a factor of N. To make BoN more efficient at inference time, one strategy is to fine-tune the language model to mimic what BoN does during inference. To achieve this, we derive the distribution induced by the BoN algorithm. We then propose to fine-tune the language model to minimize backward KL divergence to the BoN distribution. Our approach is analogous to mean-field variational inference and, thus, we term it variational BoN (vBoN). To the extent this fine-tuning is successful and we end up with a good approximation, we have reduced the inference cost by a factor of N. Our experiments on controlled generation and summarization tasks show that BoN is the most effective alignment method, and our variational approximation to BoN achieves the closest performance to BoN and surpasses models fine-tuned using the standard KL-constrained RL objective. In the controlled generation task, vBoN appears more frequently on the Pareto frontier of reward and KL divergence compared to other alignment methods. In the summarization task, vBoN achieves high reward values across various sampling temperatures.
♻ ☆ FLEXtime: Filterbank learning to explain time series
State-of-the-art methods for explaining predictions from time series involve learning an instance-wise saliency mask for each time step; however, many types of time series are difficult to interpret in the time domain, due to the inherently complex nature of the data. Instead, we propose to view time series explainability as saliency maps over interpretable parts, leaning on established signal processing methodology on signal decomposition. Specifically, we propose a new method called FLEXtime that uses a bank of bandpass filters to split the time series into frequency bands. Then, we learn the combination of these bands that optimally explains the model's prediction. Our extensive evaluation shows that, on average, FLEXtime outperforms state-of-the-art explainability methods across a range of datasets. FLEXtime fills an important gap in the current time series explainability methodology and is a valuable tool for a wide range of time series such as EEG and audio. Code will be made available at https://github.com/theabrusch/FLEXtime.
♻ ☆ Adaptive Prompt: Unlocking the Power of Visual Prompt Tuning
Visual Prompt Tuning (VPT) has recently emerged as a powerful method for adapting pre-trained vision models to downstream tasks. By introducing learnable prompt tokens as task-specific instructions, VPT effectively guides pre-trained transformer models with minimal overhead. Despite its empirical success, a comprehensive theoretical understanding of VPT remains an active area of research. Building on recent insights into the connection between mixture of experts and prompt-based approaches, we identify a key limitation in VPT: the restricted functional expressiveness in prompt formulation. To address this limitation, we propose Visual Adaptive Prompt Tuning (VAPT), a new generation of prompts that redefines prompts as adaptive functions of the input. Our theoretical analysis shows that this simple yet intuitive approach achieves optimal sample efficiency. Empirical results on VTAB-1K and FGVC further demonstrate VAPT's effectiveness, with performance gains of 7.34% and 1.04% over fully fine-tuning baselines, respectively. Notably, VAPT also surpasses VPT by a substantial margin while using fewer parameters. These results highlight both the effectiveness and efficiency of our method and pave the way for future research to explore the potential of adaptive prompts.
comment: 57 pages, 10 figures, 18 tables
♻ ☆ HOPE: A Reinforcement Learning-based Hybrid Policy Path Planner for Diverse Parking Scenarios
Automated parking stands as a highly anticipated application of autonomous driving technology. However, existing path planning methodologies fall short of addressing this need due to their incapability to handle the diverse and complex parking scenarios in reality. While non-learning methods provide reliable planning results, they are vulnerable to intricate occasions, whereas learning-based ones are good at exploration but unstable in converging to feasible solutions. To leverage the strengths of both approaches, we introduce Hybrid pOlicy Path plannEr (HOPE). This novel solution integrates a reinforcement learning agent with Reeds-Shepp curves, enabling effective planning across diverse scenarios. HOPE guides the exploration of the reinforcement learning agent by applying an action mask mechanism and employs a transformer to integrate the perceived environmental information with the mask. To facilitate the training and evaluation of the proposed planner, we propose a criterion for categorizing the difficulty level of parking scenarios based on space and obstacle distribution. Experimental results demonstrate that our approach outperforms typical rule-based algorithms and traditional reinforcement learning methods, showing higher planning success rates and generalization across various scenarios. We also conduct real-world experiments to verify the practicability of HOPE. The code for our solution is openly available on https://github.com/jiamiya/HOPE.
comment: Accepted by T-ITS. 11 pages, 5 tables, 6 figures, 2 page appendix
♻ ☆ PnP-Flow: Plug-and-Play Image Restoration with Flow Matching
In this paper, we introduce Plug-and-Play (PnP) Flow Matching, an algorithm for solving imaging inverse problems. PnP methods leverage the strength of pre-trained denoisers, often deep neural networks, by integrating them in optimization schemes. While they achieve state-of-the-art performance on various inverse problems in imaging, PnP approaches face inherent limitations on more generative tasks like inpainting. On the other hand, generative models such as Flow Matching pushed the boundary in image sampling yet lack a clear method for efficient use in image restoration. We propose to combine the PnP framework with Flow Matching (FM) by defining a time-dependent denoiser using a pre-trained FM model. Our algorithm alternates between gradient descent steps on the data-fidelity term, reprojections onto the learned FM path, and denoising. Notably, our method is computationally efficient and memory-friendly, as it avoids backpropagation through ODEs and trace computations. We evaluate its performance on denoising, super-resolution, deblurring, and inpainting tasks, demonstrating superior results compared to existing PnP algorithms and Flow Matching based state-of-the-art methods.
♻ ☆ Meta Curvature-Aware Minimization for Domain Generalization
Domain generalization (DG) aims to enhance the ability of models trained on source domains to generalize effectively to unseen domains. Recently, Sharpness-Aware Minimization (SAM) has shown promise in this area by reducing the sharpness of the loss landscape to obtain more generalized models. However, SAM and its variants sometimes fail to guide the model toward a flat minimum, and their training processes exhibit limitations, hindering further improvements in model generalization. In this paper, we first propose an improved model training process aimed at encouraging the model to converge to a flat minima. To achieve this, we design a curvature metric that has a minimal effect when the model is far from convergence but becomes increasingly influential in indicating the curvature of the minima as the model approaches a local minimum. Then we derive a novel algorithm from this metric, called Meta Curvature-Aware Minimization (MeCAM), to minimize the curvature around the local minima. Specifically, the optimization objective of MeCAM simultaneously minimizes the regular training loss, the surrogate gap of SAM, and the surrogate gap of meta-learning. We provide theoretical analysis on MeCAM's generalization error and convergence rate, and demonstrate its superiority over existing DG methods through extensive experiments on five benchmark DG datasets, including PACS, VLCS, OfficeHome, TerraIncognita, and DomainNet. Code will be available on GitHub.
comment: 22 pages, 5 figures, 17 tables
♻ ☆ Towards Training One-Step Diffusion Models Without Distillation
Recent advances in one-step generative models typically follow a two-stage process: first training a teacher diffusion model and then distilling it into a one-step student model. This distillation process traditionally relies on both the teacher model's score function to compute the distillation loss and its weights for student initialization. In this paper, we explore whether one-step generative models can be trained directly without this distillation process. First, we show that the teacher's score function is not essential and propose a family of distillation methods that achieve competitive results without relying on score estimation. Next, we demonstrate that initialization from teacher weights is indispensable in successful training. Surprisingly, we find that this benefit is not due to improved ``input-output" mapping but rather the learned feature representations, which dominate distillation quality. Our findings provide a better understanding of the role of initialization in one-step model training and its impact on distillation quality.
comment: 13 pages, Technical Report
♻ ☆ TAG: A Decentralized Framework for Multi-Agent Hierarchical Reinforcement Learning
Hierarchical organization is fundamental to biological systems and human societies, yet artificial intelligence systems often rely on monolithic architectures that limit adaptability and scalability. Current hierarchical reinforcement learning (HRL) approaches typically restrict hierarchies to two levels or require centralized training, which limits their practical applicability. We introduce TAME Agent Framework (TAG), a framework for constructing fully decentralized hierarchical multi-agent systems.TAG enables hierarchies of arbitrary depth through a novel LevelEnv concept, which abstracts each hierarchy level as the environment for the agents above it. This approach standardizes information flow between levels while preserving loose coupling, allowing for seamless integration of diverse agent types. We demonstrate the effectiveness of TAG by implementing hierarchical architectures that combine different RL agents across multiple levels, achieving improved performance over classical multi-agent RL baselines on standard benchmarks. Our results show that decentralized hierarchical organization enhances both learning speed and final performance, positioning TAG as a promising direction for scalable multi-agent systems.
♻ ☆ Slowing Down Forgetting in Continual Learning
A common challenge in continual learning (CL) is catastrophic forgetting, where the performance on old tasks drops after new, additional tasks are learned. In this paper, we propose a novel framework called ReCL to slow down forgetting in CL. Our framework exploits an implicit bias of gradient-based neural networks due to which these converge to margin maximization points. Such convergence points allow us to reconstruct old data from previous tasks, which we then combine with the current training data. Our framework is flexible and can be applied on top of existing, state-of-the-art CL methods. We further demonstrate the performance gain from our framework across a large series of experiments, including two challenging CL scenarios (class incremental and domain incremental learning), different datasets (MNIST, CIFAR10, TinyImagenet), and different network architectures. Across all experiments, we find large performance gains through ReCL. To the best of our knowledge, our framework is the first to address catastrophic forgetting by leveraging models in CL as their own memory buffers.
♻ ☆ Causality Is Key to Understand and Balance Multiple Goals in Trustworthy ML and Foundation Models
Ensuring trustworthiness in machine learning (ML) systems is crucial as they become increasingly embedded in high-stakes domains. This paper advocates for integrating causal methods into machine learning to navigate the trade-offs among key principles of trustworthy ML, including fairness, privacy, robustness, accuracy, and explainability. While these objectives should ideally be satisfied simultaneously, they are often addressed in isolation, leading to conflicts and suboptimal solutions. Drawing on existing applications of causality in ML that successfully align goals such as fairness and accuracy or privacy and robustness, this paper argues that a causal approach is essential for balancing multiple competing objectives in both trustworthy ML and foundation models. Beyond highlighting these trade-offs, we examine how causality can be practically integrated into ML and foundation models, offering solutions to enhance their reliability and interpretability. Finally, we discuss the challenges, limitations, and opportunities in adopting causal frameworks, paving the way for more accountable and ethically sound AI systems.
♻ ☆ MANTRA: The Manifold Triangulations Assemblage ICLR 2025
The rising interest in leveraging higher-order interactions present in complex systems has led to a surge in more expressive models exploiting higher-order structures in the data, especially in topological deep learning (TDL), which designs neural networks on higher-order domains such as simplicial complexes. However, progress in this field is hindered by the scarcity of datasets for benchmarking these architectures. To address this gap, we introduce MANTRA, the first large-scale, diverse, and intrinsically higher-order dataset for benchmarking higher-order models, comprising over 43,000 and 250,000 triangulations of surfaces and three-dimensional manifolds, respectively. With MANTRA, we assess several graph- and simplicial complex-based models on three topological classification tasks. We demonstrate that while simplicial complex-based neural networks generally outperform their graph-based counterparts in capturing simple topological invariants, they also struggle, suggesting a rethink of TDL. Thus, MANTRA serves as a benchmark for assessing and advancing topological methods, leading the way for more effective higher-order models.
comment: Accepted at ICLR 2025 (https://openreview.net/forum?id=X6y5CC44HM)
♻ ☆ Attacking Large Language Models with Projected Gradient Descent
Current LLM alignment methods are readily broken through specifically crafted adversarial prompts. While crafting adversarial prompts using discrete optimization is highly effective, such attacks typically use more than 100,000 LLM calls. This high computational cost makes them unsuitable for, e.g., quantitative analyses and adversarial training. To remedy this, we revisit Projected Gradient Descent (PGD) on the continuously relaxed input prompt. Although previous attempts with ordinary gradient-based attacks largely failed, we show that carefully controlling the error introduced by the continuous relaxation tremendously boosts their efficacy. Our PGD for LLMs is up to one order of magnitude faster than state-of-the-art discrete optimization to achieve the same devastating attack results.
♻ ☆ EXACFS -- A CIL Method to mitigate Catastrophic Forgetting
Deep neural networks (DNNS) excel at learning from static datasets but struggle with continual learning, where data arrives sequentially. Catastrophic forgetting, the phenomenon of forgetting previously learned knowledge, is a primary challenge. This paper introduces EXponentially Averaged Class-wise Feature Significance (EXACFS) to mitigate this issue in the class incremental learning (CIL) setting. By estimating the significance of model features for each learned class using loss gradients, gradually aging the significance through the incremental tasks and preserving the significant features through a distillation loss, EXACFS effectively balances remembering old knowledge (stability) and learning new knowledge (plasticity). Extensive experiments on CIFAR-100 and ImageNet-100 demonstrate EXACFS's superior performance in preserving stability while acquiring plasticity.
♻ ☆ Exact Certification of (Graph) Neural Networks Against Label Poisoning ICLR 2025
Machine learning models are highly vulnerable to label flipping, i.e., the adversarial modification (poisoning) of training labels to compromise performance. Thus, deriving robustness certificates is important to guarantee that test predictions remain unaffected and to understand worst-case robustness behavior. However, for Graph Neural Networks (GNNs), the problem of certifying label flipping has so far been unsolved. We change this by introducing an exact certification method, deriving both sample-wise and collective certificates. Our method leverages the Neural Tangent Kernel (NTK) to capture the training dynamics of wide networks enabling us to reformulate the bilevel optimization problem representing label flipping into a Mixed-Integer Linear Program (MILP). We apply our method to certify a broad range of GNN architectures in node classification tasks. Thereby, concerning the worst-case robustness to label flipping: $(i)$ we establish hierarchies of GNNs on different benchmark graphs; $(ii)$ quantify the effect of architectural choices such as activations, depth and skip-connections; and surprisingly, $(iii)$ uncover a novel phenomenon of the robustness plateauing for intermediate perturbation budgets across all investigated datasets and architectures. While we focus on GNNs, our certificates are applicable to sufficiently wide NNs in general through their NTK. Thus, our work presents the first exact certificate to a poisoning attack ever derived for neural networks, which could be of independent interest. The code is available at https://github.com/saper0/qpcert.
comment: Published as a spotlight presentation at ICLR 2025
♻ ☆ ReFocus: Reinforcing Mid-Frequency and Key-Frequency Modeling for Multivariate Time Series Forecasting
Recent advancements have progressively incorporated frequency-based techniques into deep learning models, leading to notable improvements in accuracy and efficiency for time series analysis tasks. However, the Mid-Frequency Spectrum Gap in the real-world time series, where the energy is concentrated at the low-frequency region while the middle-frequency band is negligible, hinders the ability of existing deep learning models to extract the crucial frequency information. Additionally, the shared Key-Frequency in multivariate time series, where different time series share indistinguishable frequency patterns, is rarely exploited by existing literature. This work introduces a novel module, Adaptive Mid-Frequency Energy Optimizer, based on convolution and residual learning, to emphasize the significance of mid-frequency bands. We also propose an Energy-based Key-Frequency Picking Block to capture shared Key-Frequency, which achieves superior inter-series modeling performance with fewer parameters. A novel Key-Frequency Enhanced Training strategy is employed to further enhance Key-Frequency modeling, where spectral information from other channels is randomly introduced into each channel. Our approach advanced multivariate time series forecasting on the challenging Traffic, ECL, and Solar benchmarks, reducing MSE by 4%, 6%, and 5% compared to the previous SOTA iTransformer. Code is available at this GitHub Repository: https://github.com/Levi-Ackman/ReFocus.
comment: Under Review
♻ ☆ Deep Learning-Driven Malware Classification with API Call Sequence Analysis and Concept Drift Handling
Malware classification in dynamic environments presents a significant challenge due to concept drift, where the statistical properties of malware data evolve over time, complicating detection efforts. To address this issue, we propose a deep learning framework enhanced with a genetic algorithm to improve malware classification accuracy and adaptability. Our approach incorporates mutation operations and fitness score evaluations within genetic algorithms to continuously refine the deep learning model, ensuring robustness against evolving malware threats. Experimental results demonstrate that this hybrid method significantly enhances classification performance and adaptability, outperforming traditional static models. Our proposed approach offers a promising solution for real-time malware classification in ever-changing cybersecurity landscapes.
♻ ☆ Q-Adapter: Customizing Pre-trained LLMs to New Preferences with Forgetting Mitigation ICLR 2025
Large Language Models (LLMs), trained on a large amount of corpus, have demonstrated remarkable abilities. However, it may not be sufficient to directly apply open-source LLMs like Llama to certain real-world scenarios, since most of them are trained for \emph{general} purposes. Thus, the demands for customizing publicly available LLMs emerge, but are currently under-studied. In this work, we consider customizing pre-trained LLMs with new human preferences. Specifically, the LLM should not only meet the new preference but also preserve its original capabilities after customization. Drawing inspiration from the observation that human preference can be expressed as a reward model, we propose to cast LLM customization as optimizing the sum of two reward functions, one of which (denoted as $r_1$) was used to pre-train the LLM while the other (denoted as $r_2$) characterizes the new human preference. The obstacle here is that both reward functions are unknown, making the application of modern reinforcement learning methods infeasible. Thanks to the residual Q-learning framework, we can restore the customized LLM with the pre-trained LLM and the \emph{residual Q-function} without the reward function $r_1$. Moreover, we find that for a fixed pre-trained LLM, the reward function $r_2$ can be derived from the residual Q-function, enabling us to directly learn the residual Q-function from the new human preference data upon the Bradley-Terry model. We name our method Q-Adapter as it introduces an adapter module to approximate the residual Q-function for customizing the pre-trained LLM towards the new preference. Experiments based on the Llama-3.1 model on the DSP dataset and HH-RLHF dataset illustrate the superior effectiveness of Q-Adapter on both retaining existing knowledge and learning new preferences. Code is available at https://github.com/mansicer/Q-Adapter.
comment: Camera ready version of ICLR 2025
♻ ☆ Scintillation pulse characterization with spectrum-inspired temporal neural networks: case studies on particle detector signals
Particle detectors based on scintillators are widely used in high-energy physics and astroparticle physics experiments, nuclear medicine imaging, industrial and environmental detection, etc. Precisely extracting scintillation signal characteristics at the event level is important for these applications, not only in respect of understanding the scintillator itself, but also kinds and physical property of incident particles. Recent researches demonstrate data-driven neural networks surpass traditional statistical methods, especially when the analytical form of signals is hard to obtain, or noise is significant. However, most densely connected or convolution-based networks fail to fully exploit the spectral and temporal structure of scintillation signals, leaving large space for performance improvement. In this paper, we propose a network architecture specially tailored for scintillation pulse characterization based on previous works on time series analysis. The core insight is that, by directly applying Fast Fourier Transform on original signals and utilizing different frequency components, the proposed network architecture can serve as a lightweight and enhanced representation learning backbone. We prove our idea in two case studies: (a) simulation data generated with the setting of the LUX dark matter detector, and (b) experimental electrical signals with fast electronics to emulate scintillation variations for the NICA/MPD calorimeter. The proposed model achieves significantly better results than the reference model in literature and densely connected models, and demonstrates higher cost-efficiency than conventional machine learning methods.
comment: 29 pages, 14 figures
♻ ☆ SURGE: On the Potential of Large Language Models as General-Purpose Surrogate Code Executors
Neural surrogate models have emerged as powerful and efficient tools in data mining. Meanwhile, large language models (LLMs) have demonstrated remarkable capabilities in code-related tasks. We investigate a novel application: using LLMs as surrogate models for code execution prediction. Given LLMs' unique ability to understand and process diverse programs, they present a promising direction for building general-purpose surrogate models. To systematically investigate this capability, we introduce SURGE, a comprehensive benchmark with $1160$ problems covering $8$ key aspects: multi-language programming tasks, competition-level programming problems, repository-level code analysis, high-cost scientific computing, time-complexity-intensive algorithms, buggy code analysis, programs dependent on specific compilers or execution environments, and formal mathematical proof verification. Through extensive empirical analysis of $21$ open-source and proprietary LLMs, we examine scaling laws, data efficiency, and predictive accuracy. Our findings reveal important insights about the feasibility of LLMs as efficient surrogates for computational processes, with implications for automated software testing, program analysis, and computational resource optimization in data mining applications. Code and dataset are released at https://github.com/Imbernoulli/SURGE.
♻ ☆ Robust Preference Optimization through Reward Model Distillation
Language model (LM) post-training (or alignment) involves maximizing a reward function that is derived from preference annotations. Direct Preference Optimization (DPO) is a popular offline alignment method that trains a policy directly on preference data without the need to train a reward model or apply reinforcement learning. However, the empirical evidence suggests that DPO typically assigns implicit rewards that overfit, and trend towards infinite magnitude. This frequently leads to degenerate policies, sometimes causing even the probabilities of the preferred generations to go to zero. In this work, we analyze this phenomenon and use distillation to get a better proxy for the true preference distribution over generation pairs: we train the LM such that its induced implicit reward, i.e., the scaled log-likelihood ratio of the model to the reference model, matches an explicit reward model trained on the preference data. Moreover, to account for uncertainty in the reward model we are distilling from, we optimize against a family of reward models that, as a whole, is likely to include at least one reasonable proxy for the preference distribution. Our results show that distilling from such a family of reward models leads to improved robustness to distribution shift in preference annotations, while preserving the simple supervised nature of DPO.
♻ ☆ Towards Graph Foundation Models: A Study on the Generalization of Positional and Structural Encodings
Recent advances in integrating positional and structural encodings (PSEs) into graph neural networks (GNNs) have significantly enhanced their performance across various graph learning tasks. However, the general applicability of these encodings and their potential to serve as foundational representations for graphs remain uncertain. This paper investigates the fine-tuning efficiency, scalability with sample size, and generalization capability of learnable PSEs across diverse graph datasets. Specifically, we evaluate their potential as universal pre-trained models that can be easily adapted to new tasks with minimal fine-tuning and limited data. Furthermore, we assess the expressivity of the learned representations, particularly, when used to augment downstream GNNs. We demonstrate through extensive benchmarking and empirical analysis that PSEs generally enhance downstream models. However, some datasets may require specific PSE-augmentations to achieve optimal performance. Nevertheless, our findings highlight their significant potential to become integral components of future graph foundation models. We provide new insights into the strengths and limitations of PSEs, contributing to the broader discourse on foundation models in graph learning.
comment: Published at TMLR (https://openreview.net/forum?id=mSoDRZXsqj)
♻ ☆ DailyDilemmas: Revealing Value Preferences of LLMs with Quandaries of Daily Life ICLR 2025
As users increasingly seek guidance from LLMs for decision-making in daily life, many of these decisions are not clear-cut and depend significantly on the personal values and ethical standards of people. We present DailyDilemmas, a dataset of 1,360 moral dilemmas encountered in everyday life. Each dilemma presents two possible actions, along with affected parties and relevant human values for each action. Based on these dilemmas, we gather a repository of human values covering diverse everyday topics, such as interpersonal relationships, workplace, and environmental issues. With DailyDilemmas, we evaluate LLMs on these dilemmas to determine what action they will choose and the values represented by these action choices. Then, we analyze values through the lens of five theoretical frameworks inspired by sociology, psychology, and philosophy, including the World Values Survey, Moral Foundations Theory, Maslow's Hierarchy of Needs, Aristotle's Virtues, and Plutchik's Wheel of Emotions. For instance, we find LLMs are most aligned with self-expression over survival in World Values Survey and care over loyalty in Moral Foundations Theory. Interestingly, we find substantial preference differences in models for some core values. For example, for truthfulness, Mixtral-8x7B neglects it by 9.7% while GPT-4-turbo selects it by 9.4%. We also study the recent guidance released by OpenAI (ModelSpec), and Anthropic (Constitutional AI) to understand how their designated principles reflect their models' actual value prioritization when facing nuanced moral reasoning in daily-life settings. Finally, we find that end users cannot effectively steer such prioritization using system prompts.
comment: Accepted into ICLR 2025 (spotlight)
♻ ☆ Test-Time Compute: from System-1 Thinking to System-2 Thinking
The remarkable performance of the o1 model in complex reasoning demonstrates that test-time compute scaling can further unlock the model's potential, enabling powerful System-2 thinking. However, there is still a lack of comprehensive surveys for test-time compute scaling. We trace the concept of test-time compute back to System-1 models. In System-1 models, test-time compute addresses distribution shifts and improves robustness and generalization through parameter updating, input modification, representation editing, and output calibration. In System-2 models, it enhances the model's reasoning ability to solve complex problems through repeated sampling, self-correction, and tree search. We organize this survey according to the trend of System-1 to System-2 thinking, highlighting the key role of test-time compute in the transition from System-1 models to weak System-2 models, and then to strong System-2 models. We also point out a few possible future directions.
comment: work in progress
♻ ☆ SheetAgent: Towards A Generalist Agent for Spreadsheet Reasoning and Manipulation via Large Language Models WWW
Spreadsheets are ubiquitous across the World Wide Web, playing a critical role in enhancing work efficiency across various domains. Large language model (LLM) has been recently attempted for automatic spreadsheet manipulation but has not yet been investigated in complicated and realistic tasks where reasoning challenges exist (e.g., long horizon manipulation with multi-step reasoning and ambiguous requirements). To bridge the gap with the real-world requirements, we introduce SheetRM, a benchmark featuring long-horizon and multi-category tasks with reasoning-dependent manipulation caused by real-life challenges. To mitigate the above challenges, we further propose SheetAgent, a novel autonomous agent that utilizes the power of LLMs. SheetAgent consists of three collaborative modules: Planner, Informer, and Retriever, achieving both advanced reasoning and accurate manipulation over spreadsheets without human interaction through iterative task reasoning and reflection. Extensive experiments demonstrate that SheetAgent delivers 20--40\% pass rate improvements on multiple benchmarks over baselines, achieving enhanced precision in spreadsheet manipulation and demonstrating superior table reasoning abilities. More details and visualizations are available at the project website: https://sheetagent.github.io/. The datasets and source code are available at https://anonymous.4open.science/r/SheetAgent.
comment: Accepted by International World Wide Web Conference (WWW) 2025 (oral)
♻ ☆ SpikeLLM: Scaling up Spiking Neural Network to Large Language Models via Saliency-based Spiking
Recent advancements in large language models (LLMs) with billions of parameters have improved performance in various applications, but their inference processes demand significant energy and computational resources. In contrast, the human brain, with approximately 86 billion neurons, is much more energy-efficient than LLMs with similar parameters. Inspired by this, we redesign 7$\sim$70 billion parameter LLMs using bio-plausible spiking mechanisms, emulating the efficient behavior of the human brain. We propose the first spiking large language model, SpikeLLM. Coupled with the proposed model, two essential approaches are proposed to improve spike training efficiency: Generalized Integrate-and-Fire (GIF) neurons to compress spike length from $T$ to $\frac{T}{L} \log_2 L$ bits, and an Optimal Brain Spiking framework to divide outlier channels and allocate different $T$ for GIF neurons, which further compresses spike length to approximate $log_2T$ bits. The necessity of spike-driven LLM is proved by comparison with quantized LLMs with similar operations. In the OmniQuant pipeline, SpikeLLM reduces 11.01% WikiText2 perplexity and improves 2.55% accuracy of common scene reasoning on a LLAMA-7B W4A4 model. In the GPTQ pipeline, SpikeLLM achieves direct additive in linear layers, significantly exceeding PB-LLMs.
♻ ☆ Efficient Learning Under Density Shift in Incremental Settings Using Cramér-Rao-Based Regularization
The continuous surge in data volume and velocity is often dealt with using data orchestration and distributed processing approaches, abstracting away the machine learning challenges that exist at the algorithmic level. With growing interest in automating the learning loop, training with data that arrive in a sequence rather than in the classical in-memory training data form will face a machine learning challenge because of evolving feature distributions across batches of training data biasing the cross-validation step (\cite{sugiyama2012machine}). This work takes a distributed density estimation angle to the problem where data are temporally distributed. It processes data in batches and allows a neural network to treat a batch as training data. The method accumulates knowledge about the data density via posterior probability absorption using the Fisher Information Matrix, which contains information about the local optimization gradients for the batch. This is then used as a regularizer for the loss in the following batch, and therefore the density estimate for the entire dataset constructively gets more robust to the non-iid distribution shift. This needs the presence of a pair of batches in memory at a time, so the space cost is not a function of the size of the complete, distributed dataset. We proposed a novel regularization-based approach Covariate Shift Correction $C^{2}A$ that leverages Fisher information and Kullback-Leibler divergence to adapt to both natural and sequential covariate shift caused by dataset fragmentation. $C^{2}A$ achieves $19\%$ accuracy at maximum against state-of-the-art methods.
comment: It is the older version of our this paper arXiv:2502.15756. So this is the duplicate older version mistakenly uploaded. There are mistakes in the method part of this paper
♻ ☆ Long-Term EEG Partitioning for Seizure Onset Detection AAAI 2025
Deep learning models have recently shown great success in classifying epileptic patients using EEG recordings. Unfortunately, classification-based methods lack a sound mechanism to detect the onset of seizure events. In this work, we propose a two-stage framework, SODor, that explicitly models seizure onset through a novel task formulation of subsequence clustering. Given an EEG sequence, the framework first learns a set of second-level embeddings with label supervision. It then employs model-based clustering to explicitly capture long-term temporal dependencies in EEG sequences and identify meaningful subsequences. Epochs within a subsequence share a common cluster assignment (normal or seizure), with cluster or state transitions representing successful onset detections. Extensive experiments on three datasets demonstrate that our method can correct misclassifications, achieving 5\%-11\% classification improvements over other baselines and accurately detecting seizure onsets.
comment: Accepted at AAAI 2025
♻ ☆ Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models ICLR 2024
Large Language Models (LLMs) have recently demonstrated remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to the large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output-channel (per-OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs. Code is available at https://github.com/johnheo/adadim-llm
comment: ICLR 2024
♻ ☆ Cross-Spectral Vision Transformer for Biometric Authentication using Forehead Subcutaneous Vein Pattern and Periocular Pattern
Traditional biometric systems have encountered significant setbacks due to various unavoidable factors, for example, face recognition-based biometrics fails due to the wearing of face masks and fingerprints create hygiene concerns. This paper proposes a novel lightweight cross-spectral vision transformer (CS-ViT) for biometric authentication using forehead subcutaneous vein patterns and periocular patterns, offering a promising alternative to traditional methods, capable of performing well even with the face masks and without any physical touch. The proposed framework comprises a cross-spectral dual-channel architecture designed to handle two distinct biometric traits and to capture inter-dependencies in terms of relative spectral patterns. Each channel consists of a Phase-Only Correlation Cross-Spectral Attention (POC-CSA) that captures their individual as well as correlated patterns. The computation of cross-spectral attention using POC extracts the phase correlation in the spatial features. Therefore, it is robust against the resolution/intensity variations and illumination of the input images, assuming both biometric traits are from the same person. The lightweight model is suitable for edge device deployment. The performance of the proposed algorithm was rigorously evaluated using the Forehead Subcutaneous Vein Pattern and Periocular Biometric Pattern (FSVP-PBP) database. The results demonstrated the superiority of the algorithm over state-of-the-art methods, achieving a remarkable classification accuracy of 98.8% with the combined vein and periocular patterns.
comment: Submitted to IEEE TPAMI
♻ ☆ Leray-Schauder Mappings for Operator Learning
We present an algorithm for learning operators between Banach spaces, based on the use of Leray-Schauder mappings to learn a finite-dimensional approximation of compact subspaces. We show that the resulting method is a universal approximator of (possibly nonlinear) operators. We demonstrate the efficiency of the approach on two benchmark datasets showing it achieves results comparable to state of the art models.
comment: 13 pages, 2 figures, 1 table. Comments are welcome! v2: Theoretical analysis expanded, several explanations regarding the experiments have been added for improved clarity
♻ ☆ Representation Engineering: A Top-Down Approach to AI Transparency
In this paper, we identify and characterize the emerging area of representation engineering (RepE), an approach to enhancing the transparency of AI systems that draws on insights from cognitive neuroscience. RepE places population-level representations, rather than neurons or circuits, at the center of analysis, equipping us with novel methods for monitoring and manipulating high-level cognitive phenomena in deep neural networks (DNNs). We provide baselines and an initial analysis of RepE techniques, showing that they offer simple yet effective solutions for improving our understanding and control of large language models. We showcase how these methods can provide traction on a wide range of safety-relevant problems, including honesty, harmlessness, power-seeking, and more, demonstrating the promise of top-down transparency research. We hope that this work catalyzes further exploration of RepE and fosters advancements in the transparency and safety of AI systems.
comment: Code is available at https://github.com/andyzoujm/representation-engineering
♻ ☆ TokenSelect: Efficient Long-Context Inference and Length Extrapolation for LLMs via Dynamic Token-Level KV Cache Selection
The rapid advancement of Large Language Models (LLMs) has driven growing demand for processing extended context sequences in contemporary applications. However, this progress faces two major challenges: performance degradation due to sequence lengths out-of-distribution, and excessively long inference times caused by the quadratic computational complexity of attention. These issues hinder the application of LLMs in long-context scenarios. In this paper, we propose Dynamic Token-Level KV Cache Selection (TokenSelect), a training-free method for efficient and accurate long-context inference. TokenSelect builds upon the observation of non-contiguous attention sparsity, using Query-Key dot products to measure per-head KV Cache criticality at token-level. By per-head soft voting mechanism, TokenSelect selectively involves a few critical KV cache tokens in attention calculation without sacrificing accuracy. To further accelerate TokenSelect, we design the Selection Cache based on observations of consecutive Query similarity and implemented efficient dot product kernel, significantly reducing the overhead. A comprehensive evaluation of TokenSelect demonstrates up to 23.84x speedup in attention computation and up to 2.28x acceleration in end-to-end latency, while providing superior performance compared to state-of-the-art long-context inference methods.
♻ ☆ Speed-accuracy relations for the diffusion models: Wisdom from nonequilibrium thermodynamics and optimal transport
We discuss a connection between a generative model, called the diffusion model, and nonequilibrium thermodynamics for the Fokker-Planck equation, called stochastic thermodynamics. Based on the techniques of stochastic thermodynamics, we derive the speed-accuracy relations for the diffusion models, which are inequalities that relate the accuracy of data generation to the entropy production rate, which can be interpreted as the speed of the diffusion dynamics in the absence of the non-conservative force. From a stochastic thermodynamic perspective, our results provide a quantitative insight into how best to generate data in diffusion models. The optimal learning protocol is introduced by the geodesic of space of the 2-Wasserstein distance in optimal transport theory. We numerically illustrate the validity of the speed-accuracy relations for the diffusion models with different noise schedules and the different data. We numerically discuss our results for the optimal and suboptimal learning protocols. We also show the inaccurate data generation due to the non-conservative force, and the applicability of our results to data generation from the real-world image datasets.
comment: 36 pages, 7 figures
♻ ☆ Structural-Entropy-Based Sample Selection for Efficient and Effective Learning ICLR 2025
Sample selection improves the efficiency and effectiveness of machine learning models by providing informative and representative samples. Typically, samples can be modeled as a sample graph, where nodes are samples and edges represent their similarities. Most existing methods are based on local information, such as the training difficulty of samples, thereby overlooking global information, such as connectivity patterns. This oversight can result in suboptimal selection because global information is crucial for ensuring that the selected samples well represent the structural properties of the graph. To address this issue, we employ structural entropy to quantify global information and losslessly decompose it from the whole graph to individual nodes using the Shapley value. Based on the decomposition, we present $\textbf{S}$tructural-$\textbf{E}$ntropy-based sample $\textbf{S}$election ($\textbf{SES}$), a method that integrates both global and local information to select informative and representative samples. SES begins by constructing a $k$NN-graph among samples based on their similarities. It then measures sample importance by combining structural entropy (global metric) with training difficulty (local metric). Finally, SES applies importance-biased blue noise sampling to select a set of diverse and representative samples. Comprehensive experiments on three learning scenarios -- supervised learning, active learning, and continual learning -- clearly demonstrate the effectiveness of our method.
comment: Published as a conference paper at ICLR 2025
♻ ☆ PATCH: a deep learning method to assess heterogeneity of artistic practice in historical paintings
The history of art has seen significant shifts in the manner in which artworks are created, making understanding of creative processes a central question in technical art history. In the Renaissance and Early Modern period, paintings were largely produced by master painters directing workshops of apprentices who often contributed to projects. The masters varied significantly in artistic and managerial styles, meaning different combinations of artists and implements might be seen both between masters and within workshops or even individual canvases. Information on how different workshops were managed and the processes by which artworks were created remains elusive. Machine learning methods have potential to unearth new information about artists' creative processes by extending the analysis of brushwork to a microscopic scale. Analysis of workshop paintings, however, presents a challenge in that documentation of the artists and materials involved is sparse, meaning external examples are not available to train networks to recognize their contributions. Here we present a novel machine learning approach we call pairwise assignment training for classifying heterogeneity (PATCH) that is capable of identifying individual artistic practice regimes with no external training data, or "ground truth." The method achieves unsupervised results by supervised means, and outperforms both simple statistical procedures and unsupervised machine learning methods. We apply this method to two historical paintings by the Spanish Renaissance master, El Greco: The Baptism of Christ and Christ on the Cross with Landscape, and our findings regarding the former potentially challenge previous work that has assigned the painting to workshop members. Further, the results of our analyses create a measure of heterogeneity of artistic practice that can be used to characterize artworks across time and space.
comment: main text: 16 pages, 6 figures; SI: 7 pages, 3 figures; v2: minor typo corrections, higher resolution figures
♻ ☆ Node-Time Conditional Prompt Learning In Dynamic Graphs ICLR 2025
Dynamic graphs capture evolving interactions between entities, such as in social networks, online learning platforms, and crowdsourcing projects. For dynamic graph modeling, dynamic graph neural networks (DGNNs) have emerged as a mainstream technique. However, they are generally pre-trained on the link prediction task, leaving a significant gap from the objectives of downstream tasks such as node classification. To bridge the gap, prompt-based learning has gained traction on graphs, but most existing efforts focus on static graphs, neglecting the evolution of dynamic graphs. In this paper, we propose DYGPROMPT, a novel pre-training and prompt learning framework for dynamic graph modeling. First, we design dual prompts to address the gap in both task objectives and temporal variations across pre-training and downstream tasks. Second, we recognize that node and time features mutually characterize each other, and propose dual condition-nets to model the evolving node-time patterns in downstream tasks. Finally, we thoroughly evaluate and analyze DYGPROMPT through extensive experiments on four public datasets.
comment: Accepted by ICLR 2025
♻ ☆ AdvLogo: Adversarial Patch Attack against Object Detectors based on Diffusion Models
With the rapid development of deep learning, object detectors have demonstrated impressive performance; however, vulnerabilities still exist in certain scenarios. Current research exploring the vulnerabilities using adversarial patches often struggles to balance the trade-off between attack effectiveness and visual quality. To address this problem, we propose a novel framework of patch attack from semantic perspective, which we refer to as AdvLogo. Based on the hypothesis that every semantic space contains an adversarial subspace where images can cause detectors to fail in recognizing objects, we leverage the semantic understanding of the diffusion denoising process and drive the process to adversarial subareas by perturbing the latent and unconditional embeddings at the last timestep. To mitigate the distribution shift that exposes a negative impact on image quality, we apply perturbation to the latent in frequency domain with the Fourier Transform. Experimental results demonstrate that AdvLogo achieves strong attack performance while maintaining high visual quality.
♻ ☆ Generative Representational Instruction Tuning
All text-based language problems can be reduced to either generation or embedding. Current models only perform well at one or the other. We introduce generative representational instruction tuning (GRIT) whereby a large language model is trained to handle both generative and embedding tasks by distinguishing between them through instructions. Compared to other open models, our resulting GritLM 7B sets a new state of the art on the Massive Text Embedding Benchmark (MTEB) and outperforms all models up to its size on a range of generative tasks. By scaling up further, GritLM 8x7B outperforms all open generative language models that we tried while still being among the best embedding models. Notably, we find that GRIT matches training on only generative or embedding data, thus we can unify both at no performance loss. Among other benefits, the unification via GRIT speeds up Retrieval-Augmented Generation (RAG) by > 60% for long documents, by no longer requiring separate retrieval and generation models. Models, code, etc. are freely available at https://github.com/ContextualAI/gritlm.
comment: 67 pages (16 main), 25 figures, 34 tables
♻ ☆ Calib3D: Calibrating Model Preferences for Reliable 3D Scene Understanding WACV 2025
Safety-critical 3D scene understanding tasks necessitate not only accurate but also confident predictions from 3D perception models. This study introduces Calib3D, a pioneering effort to benchmark and scrutinize the reliability of 3D scene understanding models from an uncertainty estimation viewpoint. We comprehensively evaluate 28 state-of-the-art models across 10 diverse 3D datasets, uncovering insightful phenomena that cope with both the aleatoric and epistemic uncertainties in 3D scene understanding. We discover that despite achieving impressive levels of accuracy, existing models frequently fail to provide reliable uncertainty estimates -- a pitfall that critically undermines their applicability in safety-sensitive contexts. Through extensive analysis of key factors such as network capacity, LiDAR representations, rasterization resolutions, and 3D data augmentation techniques, we correlate these aspects directly with the model calibration efficacy. Furthermore, we introduce DeptS, a novel depth-aware scaling approach aimed at enhancing 3D model calibration. Extensive experiments across a wide range of configurations validate the superiority of our method. We hope this work could serve as a cornerstone for fostering reliable 3D scene understanding. Code and benchmark toolkit are publicly available.
comment: WACV 2025 Oral; 26 pages, 8 figures, 12 tables; Code at https://github.com/ldkong1205/Calib3D
♻ ☆ Federated Learning in Practice: Reflections and Projections
Federated Learning (FL) is a machine learning technique that enables multiple entities to collaboratively learn a shared model without exchanging their local data. Over the past decade, FL systems have achieved substantial progress, scaling to millions of devices across various learning domains while offering meaningful differential privacy (DP) guarantees. Production systems from organizations like Google, Apple, and Meta demonstrate the real-world applicability of FL. However, key challenges remain, including verifying server-side DP guarantees and coordinating training across heterogeneous devices, limiting broader adoption. Additionally, emerging trends such as large (multi-modal) models and blurred lines between training, inference, and personalization challenge traditional FL frameworks. In response, we propose a redefined FL framework that prioritizes privacy principles rather than rigid definitions. We also chart a path forward by leveraging trusted execution environments and open-source ecosystems to address these challenges and facilitate future advancements in FL.
comment: Published at 2024 IEEE 6th International Conference on Trust, Privacy and Security in Intelligent Systems, and Applications (TPS-ISA)
♻ ☆ Compositional simulation-based inference for time series ICLR 2025
Amortized simulation-based inference (SBI) methods train neural networks on simulated data to perform Bayesian inference. While this strategy avoids the need for tractable likelihoods, it often requires a large number of simulations and has been challenging to scale to time series data. Scientific simulators frequently emulate real-world dynamics through thousands of single-state transitions over time. We propose an SBI approach that can exploit such Markovian simulators by locally identifying parameters consistent with individual state transitions. We then compose these local results to obtain a posterior over parameters that align with the entire time series observation. We focus on applying this approach to neural posterior score estimation but also show how it can be applied, e.g., to neural likelihood (ratio) estimation. We demonstrate that our approach is more simulation-efficient than directly estimating the global posterior on several synthetic benchmark tasks and simulators used in ecology and epidemiology. Finally, we validate scalability and simulation efficiency of our approach by applying it to a high-dimensional Kolmogorov flow simulator with around one million data dimensions.
comment: To be published in the proceedings of the Thirteenth International Conference on Learning Representations (ICLR 2025), Singapore, 2025
♻ ☆ Direct Distributional Optimization for Provable Alignment of Diffusion Models
We introduce a novel alignment method for diffusion models from distribution optimization perspectives while providing rigorous convergence guarantees. We first formulate the problem as a generic regularized loss minimization over probability distributions and directly optimize the distribution using the Dual Averaging method. Next, we enable sampling from the learned distribution by approximating its score function via Doob's $h$-transform technique. The proposed framework is supported by rigorous convergence guarantees and an end-to-end bound on the sampling error, which imply that when the original distribution's score is known accurately, the complexity of sampling from shifted distributions is independent of isoperimetric conditions. This framework is broadly applicable to general distribution optimization problems, including alignment tasks in Reinforcement Learning with Human Feedback (RLHF), Direct Preference Optimization (DPO), and Kahneman-Tversky Optimization (KTO). We empirically validate its performance on synthetic and image datasets using the DPO objective.
♻ ☆ Iterative Nash Policy Optimization: Aligning LLMs with General Preferences via No-Regret Learning
Reinforcement Learning with Human Feedback (RLHF) has achieved great success in aligning large language models (LLMs) with human preferences. Prevalent RLHF approaches are reward-based, following the Bradley-Terry (BT) model assumption, which may not fully capture the complexity of human preferences. In this paper, we explore RLHF under a general preference framework and approach it from a game-theoretic perspective. Specifically, we formulate the problem as a two-player game and propose a novel online algorithm, iterative Nash policy optimization (INPO). The key idea is to let the policy play against itself via no-regret learning, thereby approximating the Nash policy. Unlike previous methods, INPO bypasses the need for estimating the expected win rate for individual responses, which typically incurs high computational or annotation costs. Instead, we introduce a new loss objective that is directly minimized over a preference dataset. We provide theoretical analysis for our approach and demonstrate its effectiveness through experiments on various representative benchmarks. With an LLaMA-3-8B-based SFT model, INPO achieves a 42.6% length-controlled win rate on AlpacaEval 2.0 and a 37.8% win rate on Arena-Hard, showing substantial improvement over the state-of-the-art online RLHF algorithms.
♻ ☆ Discovering physical laws with parallel combinatorial tree search
Symbolic regression plays a crucial role in modern scientific research thanks to its capability of discovering concise and interpretable mathematical expressions from data. A grand challenge lies in the arduous search for parsimonious and generalizable mathematical formulas, in an infinite search space, while intending to fit the training data. Existing algorithms have faced a critical bottleneck of accuracy and efficiency over a decade when handling problems of complexity, which essentially hinders the pace of applying symbolic regression for scientific exploration across interdisciplinary domains. To this end, we introduce a parallel combinatorial tree search (PCTS) model to efficiently distill generic mathematical expressions from limited data. Through a series of extensive experiments, we demonstrate the superior accuracy and efficiency of PCTS for equation discovery, which greatly outperforms the state-of-the-art baseline models on over 200 synthetic and experimental datasets (e.g., lifting its performance by up to 99% accuracy improvement and one-order of magnitude speed up). PCTS represents a key advance in accurate and efficient data-driven discovery of symbolic, interpretable models (e.g., underlying physical laws) and marks a pivotal transition towards scalable symbolic learning.
♻ ☆ Learning to Learn Weight Generation via Trajectory Diffusion
Diffusion-based algorithms have emerged as promising techniques for weight generation, particularly in scenarios like multi-task learning that require frequent weight updates. However, existing solutions suffer from limited cross-task transferability. In addition, they only utilize optimal weights as training samples, ignoring the value of other weights in the optimization process. To address these issues, we propose Lt-Di, which integrates the diffusion algorithm with meta-learning to generate weights for unseen tasks. Furthermore, we extend the vanilla diffusion algorithm into a trajectory diffusion algorithm to utilize other weights along the optimization trajectory. Trajectory diffusion decomposes the entire diffusion chain into multiple shorter ones, improving training and inference efficiency. We analyze the convergence properties of the weight generation paradigm and improve convergence efficiency without additional time overhead. Our experiments demonstrate Lt-Di's higher accuracy while reducing computational overhead across various tasks, including zero-shot and few-shot learning, multi-domain generalization, and large-scale language model fine-tuning.Our code is released at https://anonymous.4open.science/r/Lt-Di-0E51.
♻ ☆ Struc2mapGAN: improving synthetic cryo-EM density maps with generative adversarial networks
Generating synthetic cryogenic electron microscopy 3D density maps from molecular structures has potential important applications in structural biology. Yet existing simulation-based methods cannot mimic all the complex features present in experimental maps, such as secondary structure elements. As an alternative, we propose struc2mapGAN, a novel data-driven method that employs a generative adversarial network to produce improved experimental-like density maps from molecular structures. More specifically, struc2mapGAN uses a nested U-Net architecture as the generator, with an additional L1 loss term and further processing of raw training experimental maps to enhance learning efficiency. While struc2mapGAN can promptly generate maps after training, we demonstrate that it outperforms existing simulation-based methods for a wide array of tested maps and across various evaluation metrics.
♻ ☆ LLMOPT: Learning to Define and Solve General Optimization Problems from Scratch
Optimization problems are prevalent across various scenarios. Formulating and then solving optimization problems described by natural language often requires highly specialized human expertise, which could block the widespread application of optimization-based decision making. To automate problem formulation and solving, leveraging large language models (LLMs) has emerged as a potential way. However, this kind of approach suffers from the issue of optimization generalization. Namely, the accuracy of most current LLM-based methods and the generality of optimization problem types that they can model are still limited. In this paper, we propose a unified learning-based framework called LLMOPT to boost optimization generalization. Starting from the natural language descriptions of optimization problems and a pre-trained LLM, LLMOPT constructs the introduced five-element formulation as a universal model for learning to define diverse optimization problem types. Then, LLMOPT employs the multi-instruction tuning to enhance both problem formalization and solver code generation accuracy and generality. After that, to prevent hallucinations in LLMs, such as sacrificing solving accuracy to avoid execution errors, the model alignment and self-correction mechanism are adopted in LLMOPT. We evaluate the optimization generalization ability of LLMOPT and compared methods across six real-world datasets covering roughly 20 fields such as health, environment, energy and manufacturing, etc. Extensive experiment results show that LLMOPT is able to model various optimization problem types such as linear/nonlinear programming, mixed integer programming, and combinatorial optimization, and achieves a notable 11.08% average solving accuracy improvement compared with the state-of-the-art methods. The code is available at https://github.com/caigaojiang/LLMOPT.
♻ ☆ TSVD: Bridging Theory and Practice in Continual Learning with Pre-trained Models ICLR 2025
The goal of continual learning (CL) is to train a model that can solve multiple tasks presented sequentially. Recent CL approaches have achieved strong performance by leveraging large pre-trained models that generalize well to downstream tasks. However, such methods lack theoretical guarantees, making them prone to unexpected failures. Conversely, principled CL approaches often fail to achieve competitive performance. In this work, we aim to bridge this gap between theory and practice by designing a simple CL method that is theoretically sound and highly performant. Specifically, we lift pre-trained features into a higher dimensional space and formulate an over-parametrized minimum-norm least-squares problem. We find that the lifted features are highly ill-conditioned, potentially leading to large training errors (numerical instability) and increased generalization errors. We address these challenges by continually truncating the singular value decomposition (SVD) of the lifted features. Our approach, termed TSVD, is stable with respect to the choice of hyperparameters, can handle hundreds of tasks, and outperforms state-of-the-art CL methods on multiple datasets. Importantly, our method satisfies a recurrence relation throughout its continual learning process, which allows us to prove it maintains small training and generalization errors by appropriately truncating a fraction of SVD factors. This results in a stable continual learning method with strong empirical performance and theoretical guarantees. Code available: https://github.com/liangzu/tsvd.
comment: 47 pages, 18 figures, 16 tables (v2, accepted to ICLR 2025)
♻ ☆ MMed-RAG: Versatile Multimodal RAG System for Medical Vision Language Models ICLR 2025
Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection method, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in the factual accuracy of Med-LVLMs. Our data and code are available in https://github.com/richard-peng-xia/MMed-RAG.
comment: ICLR 2025
♻ ☆ HORAE: A Domain-Agnostic Modeling Language for Automating Multimodal Service Regulation
Artificial intelligence is rapidly encroaching on the field of service regulation. This work-in-progress article presents the design principles behind HORAE, a unified specification language to model multimodal regulation rules across a diverse set of domains. We show how HORAE facilitates an intelligent service regulation pipeline by further exploiting a fine-tuned large language model named HORAE that automates the HORAE modeling process, thereby yielding an end-to-end framework for fully automated intelligent service regulation.
♻ ☆ Scaling Offline Model-Based RL via Jointly-Optimized World-Action Model Pretraining ICLR 2025
A significant aspiration of offline reinforcement learning (RL) is to develop a generalist agent with high capabilities from large and heterogeneous datasets. However, prior approaches that scale offline RL either rely heavily on expert trajectories or struggle to generalize to diverse unseen tasks. Inspired by the excellent generalization of world model in conditional video generation, we explore the potential of image observation-based world model for scaling offline RL and enhancing generalization on novel tasks. In this paper, we introduce JOWA: Jointly-Optimized World-Action model, an offline model-based RL agent pretrained on multiple Atari games with 6 billion tokens data to learn general-purpose representation and decision-making ability. Our method jointly optimizes a world-action model through a shared transformer backbone, which stabilize temporal difference learning with large models during pretraining. Moreover, we propose a provably efficient and parallelizable planning algorithm to compensate for the Q-value estimation error and thus search out better policies. Experimental results indicate that our largest agent, with 150 million parameters, achieves 78.9% human-level performance on pretrained games using only 10% subsampled offline data, outperforming existing state-of-the-art large-scale offline RL baselines by 31.6% on averange. Furthermore, JOWA scales favorably with model capacity and can sample-efficiently transfer to novel games using only 5k offline fine-tuning data (approximately 4 trajectories) per game, demonstrating superior generalization. We will release codes and model weights at https://github.com/CJReinforce/JOWA
comment: Accepted by ICLR 2025
♻ ☆ PhyMPGN: Physics-encoded Message Passing Graph Network for spatiotemporal PDE systems
Solving partial differential equations (PDEs) serves as a cornerstone for modeling complex dynamical systems. Recent progresses have demonstrated grand benefits of data-driven neural-based models for predicting spatiotemporal dynamics (e.g., tremendous speedup gain compared with classical numerical methods). However, most existing neural models rely on rich training data, have limited extrapolation and generalization abilities, and suffer to produce precise or reliable physical prediction under intricate conditions (e.g., irregular mesh or geometry, complex boundary conditions, diverse PDE parameters, etc.). To this end, we propose a new graph learning approach, namely, Physics-encoded Message Passing Graph Network (PhyMPGN), to model spatiotemporal PDE systems on irregular meshes given small training datasets. Specifically, we incorporate a GNN into a numerical integrator to approximate the temporal marching of spatiotemporal dynamics for a given PDE system. Considering that many physical phenomena are governed by diffusion processes, we further design a learnable Laplace block, which encodes the discrete Laplace-Beltrami operator, to aid and guide the GNN learning in a physically feasible solution space. A boundary condition padding strategy is also designed to improve the model convergence and accuracy. Extensive experiments demonstrate that PhyMPGN is capable of accurately predicting various types of spatiotemporal dynamics on coarse unstructured meshes, consistently achieves the state-of-the-art results, and outperforms other baselines with considerable gains.
♻ ☆ A Closer Look at Machine Unlearning for Large Language Models ICLR 2025
Large language models (LLMs) may memorize sensitive or copyrighted content, raising privacy and legal concerns. Due to the high cost of retraining from scratch, researchers attempt to employ machine unlearning to remove specific content from LLMs while preserving the overall performance. In this paper, we discuss several issues in machine unlearning for LLMs and provide our insights on possible approaches. To address the issue of inadequate evaluation of model outputs after unlearning, we introduce three additional metrics to evaluate token diversity, sentence semantics, and factual correctness. We then categorize unlearning methods into untargeted and targeted, and discuss their issues respectively. Specifically, the behavior that untargeted unlearning attempts to approximate is unpredictable and may involve hallucinations, and existing regularization is insufficient for targeted unlearning. To alleviate these issues, we propose using the objective of maximizing entropy (ME) for untargeted unlearning and incorporate answer preservation (AP) loss as regularization for targeted unlearning. Experimental results across three scenarios, i.e., fictitious unlearning, continual unlearning, and real-world unlearning, demonstrate the effectiveness of our approaches. The code is available at https://github.com/sail-sg/closer-look-LLM-unlearning.
comment: ICLR 2025
♻ ☆ A Lean Dataset for International Math Olympiad: Small Steps towards Writing Math Proofs for Hard Problems
Using AI to write formal proofs for mathematical problems is a challenging task that has seen some advancements in recent years. Automated systems such as Lean can verify the correctness of proofs written in formal language, yet writing the proofs in formal language can be challenging for humans and machines. The miniF2F benchmark has 20 IMO problems in its test set, yet formal proofs are available only for 6 of these problems (3 of which are only written by mathematicians). The model with best accuracy can only prove 2 of these 20 IMO problems, from 1950s and 60s, while its training set is a secret. In this work, we write complete, original formal proofs for the remaining IMO problems in Lean along with 3 extra problems from IMO 2022 and 2023. This effort expands the availability of proof currently in the public domain by creating 5,880 lines of Lean proof. The goal of the paper is to pave the way for developing AI models that can automatically write the formal proofs for all the IMO problems in miniF2F and beyond by providing an evaluation benchmark. In this pursuit, we devise a method to decompose the proofs of these problems into their building blocks, constructing a dataset of 1,329 lemmas with more than 40k lines of Lean code. These lemmas are not trivial, yet they are approachable, providing the opportunity to evaluate and diagnose the failures and successes of AI models. We evaluate the ability of the SOTA LLMs on our dataset and analyze their success and failure modes from different perspectives. Our dataset and code is available at: https://github.com/roozbeh-yz/IMO-Steps.
♻ ☆ Scalable Decision-Making in Stochastic Environments through Learned Temporal Abstraction ICLR2025
Sequential decision-making in high-dimensional continuous action spaces, particularly in stochastic environments, faces significant computational challenges. We explore this challenge in the traditional offline RL setting, where an agent must learn how to make decisions based on data collected through a stochastic behavior policy. We present Latent Macro Action Planner (L-MAP), which addresses this challenge by learning a set of temporally extended macro-actions through a state-conditional Vector Quantized Variational Autoencoder (VQ-VAE), effectively reducing action dimensionality. L-MAP employs a (separate) learned prior model that acts as a latent transition model and allows efficient sampling of plausible actions. During planning, our approach accounts for stochasticity in both the environment and the behavior policy by using Monte Carlo tree search (MCTS). In offline RL settings, including stochastic continuous control tasks, L-MAP efficiently searches over discrete latent actions to yield high expected returns. Empirical results demonstrate that L-MAP maintains low decision latency despite increased action dimensionality. Notably, across tasks ranging from continuous control with inherently stochastic dynamics to high-dimensional robotic hand manipulation, L-MAP significantly outperforms existing model-based methods and performs on-par with strong model-free actor-critic baselines, highlighting the effectiveness of the proposed approach in planning in complex and stochastic environments with high-dimensional action spaces.
comment: Accepted by ICLR2025. Code would be available at https://github.com/BaitingLuo/L-MAP.git
♻ ☆ On the Feature Learning in Diffusion Models
The predominant success of diffusion models in generative modeling has spurred significant interest in understanding their theoretical foundations. In this work, we propose a feature learning framework aimed at analyzing and comparing the training dynamics of diffusion models with those of traditional classification models. Our theoretical analysis demonstrates that diffusion models, due to the denoising objective, are encouraged to learn more balanced and comprehensive representations of the data. In contrast, neural networks with a similar architecture trained for classification tend to prioritize learning specific patterns in the data, often focusing on easy-to-learn components. To support these theoretical insights, we conduct several experiments on both synthetic and real-world datasets, which empirically validate our findings and highlight the distinct feature learning dynamics in diffusion models compared to classification.
♻ ☆ Weighted Point Set Embedding for Multimodal Contrastive Learning Toward Optimal Similarity Metric ICLR 2025
In typical multimodal contrastive learning, such as CLIP, encoders produce one point in the latent representation space for each input. However, one-point representation has difficulty in capturing the relationship and the similarity structure of a huge amount of instances in the real world. For richer classes of the similarity, we propose the use of weighted point sets, namely, sets of pairs of weight and vector, as representations of instances. In this work, we theoretically show the benefit of our proposed method through a new understanding of the contrastive loss of CLIP, which we call symmetric InfoNCE. We clarify that the optimal similarity that minimizes symmetric InfoNCE is the pointwise mutual information, and show an upper bound of excess risk on downstream classification tasks of representations that achieve the optimal similarity. In addition, we show that our proposed similarity based on weighted point sets consistently achieves the optimal similarity. To verify the effectiveness of our proposed method, we demonstrate pretraining of text-image representation models and classification tasks on common benchmarks.
comment: ICLR 2025 (Spotlight)
♻ ☆ Penalized Principal Component Analysis Using Smoothing
Principal components computed via PCA (principal component analysis) are traditionally used to reduce dimensionality in genomic data or to correct for population stratification. In this paper, we explore the penalized eigenvalue problem (PEP) which reformulates the computation of the first eigenvector as an optimization problem and adds an $L_1$ penalty constraint to enforce sparseness of the solution. The contribution of our article is threefold. First, we extend PEP by applying smoothing to the original LASSO-type $L_1$ penalty. This allows one to compute analytical gradients which enable faster and more efficient minimization of the objective function associated with the optimization problem. Second, we demonstrate how higher order eigenvectors can be calculated with PEP using established results from singular value decomposition (SVD). Third, we present four experimental studies to demonstrate the usefulness of the smoothed penalized eigenvectors. Using data from the 1000 Genomes Project dataset, we empirically demonstrate that our proposed smoothed PEP allows one to increase numerical stability and obtain meaningful eigenvectors. We also employ the penalized eigenvector approach in two additional real data applications (computation of a polygenic risk score and clustering), demonstrating that exchanging the penalized eigenvectors for their smoothed counterparts can increase prediction accuracy in polygenic risk scores and enhance discernibility of clusterings. Moreover, we compare our proposed smoothed PEP to seven state-of-the-art algorithms for sparse PCA and evaluate the accuracy of the obtained eigenvectors, their support recovery, and their runtime.
♻ ☆ OLMoE: Open Mixture-of-Experts Language Models
We introduce OLMoE, a fully open, state-of-the-art language model leveraging sparse Mixture-of-Experts (MoE). OLMoE-1B-7B has 7 billion (B) parameters but uses only 1B per input token. We pretrain it on 5 trillion tokens and further adapt it to create OLMoE-1B-7B-Instruct. Our models outperform all available models with similar active parameters, even surpassing larger ones like Llama2-13B-Chat and DeepSeekMoE-16B. We present various experiments on MoE training, analyze routing in our model showing high specialization, and open-source all aspects of our work: model weights, training data, code, and logs.
comment: 63 pages (24 main), 36 figures, 17 tables
♻ ☆ On Large Language Model Continual Unlearning ICLR 2025
While large language models have demonstrated impressive performance across various domains and tasks, their security issues have become increasingly severe. Machine unlearning has emerged as a representative approach for model safety and security by removing the influence of undesired data on the target model. However, these methods do not sufficiently consider that unlearning requests in real-world scenarios are continuously emerging, especially in the context of LLMs, which may lead to accumulated model utility loss that eventually becomes unacceptable. Moreover, existing LLM unlearning methods often ignore previous data access limitations due to privacy concerns and copyright protection. Without previous data, the utility preservation during unlearning is much harder. To overcome these challenges, we propose the OOO framework that includes an Orthogonal low-rank adapter (LoRA) for continually unlearning requested data and an Out-Of-Distribution (OOD) detector to measure the similarity between input and unlearning data. The orthogonal LoRA achieves parameter disentanglement among continual unlearning requests. The OOD detector is trained with a novel contrastive entropy loss and utilizes a glocal-aware scoring mechanism. During inference, our OOO framework can decide whether and to what extent to load the unlearning LoRA based on the OOD detector's predicted similarity between the input and the unlearned knowledge. Notably, OOO's effectiveness does not rely on any retained data. We conducted extensive experiments on OOO and state-of-the-art LLM unlearning methods across three tasks and seven datasets. The results indicate that OOO consistently achieves the best unlearning effectiveness and utility preservation, especially when facing continuous unlearning requests. The source codes can be found at https://github.com/GCYZSL/O3-LLM-UNLEARNING.
comment: This paper has been accepted by ICLR 2025. The first two authors contribute equally and they are ordered alphabetically
♻ ☆ BECAUSE: Bilinear Causal Representation for Generalizable Offline Model-based Reinforcement Learning
Offline model-based reinforcement learning (MBRL) enhances data efficiency by utilizing pre-collected datasets to learn models and policies, especially in scenarios where exploration is costly or infeasible. Nevertheless, its performance often suffers from the objective mismatch between model and policy learning, resulting in inferior performance despite accurate model predictions. This paper first identifies the primary source of this mismatch comes from the underlying confounders present in offline data for MBRL. Subsequently, we introduce \textbf{B}ilin\textbf{E}ar \textbf{CAUS}al r\textbf{E}presentation~(BECAUSE), an algorithm to capture causal representation for both states and actions to reduce the influence of the distribution shift, thus mitigating the objective mismatch problem. Comprehensive evaluations on 18 tasks that vary in data quality and environment context demonstrate the superior performance of BECAUSE over existing offline RL algorithms. We show the generalizability and robustness of BECAUSE under fewer samples or larger numbers of confounders. Additionally, we offer theoretical analysis of BECAUSE to prove its error bound and sample efficiency when integrating causal representation into offline MBRL.
♻ ☆ RobotFingerPrint: Unified Gripper Coordinate Space for Multi-Gripper Grasp Synthesis and Transfer
We introduce a novel grasp representation named the Unified Gripper Coordinate Space (UGCS) for grasp synthesis and grasp transfer. Our representation leverages spherical coordinates to create a shared coordinate space across different robot grippers, enabling it to synthesize and transfer grasps for both novel objects and previously unseen grippers. The strength of this representation lies in the ability to map palm and fingers of a gripper and the unified coordinate space. Grasp synthesis is formulated as predicting the unified spherical coordinates on object surface points via a conditional variational autoencoder. The predicted unified gripper coordinates establish exact correspondences between the gripper and object points, which is used to optimize grasp pose and joint values. Grasp transfer is facilitated through the point-to-point correspondence between any two (potentially unseen) grippers and solved via a similar optimization. Extensive simulation and real-world experiments showcase the efficacy of the unified grasp representation for grasp synthesis in generating stable and diverse grasps. Similarly, we showcase real-world grasp transfer from human demonstrations across different objects.
comment: 8 pages, 11 figures, 3 tables. Project page available at https://irvlutd.github.io/RobotFingerPrint
♻ ☆ The Labyrinth of Links: Navigating the Associative Maze of Multi-modal LLMs ICLR 2025
Multi-modal Large Language Models (MLLMs) have exhibited impressive capability. However, recently many deficiencies of MLLMs have been found compared to human intelligence, $\textit{e.g.}$, hallucination. To drive the MLLMs study, the community dedicated efforts to building larger benchmarks with complex tasks. In this paper, we propose benchmarking an essential but usually overlooked intelligence: $\textbf{association}$, a human's basic capability to link observation and prior practice memory. To comprehensively investigate MLLM's performance on the association, we formulate the association task and devise a standard benchmark based on adjective and verb semantic concepts. Instead of costly data annotation and curation, we propose a convenient $\textbf{annotation-free}$ construction method transforming the general dataset for our association tasks. Simultaneously, we devise a rigorous data refinement process to eliminate confusion in the raw dataset. Building on this database, we establish three levels of association tasks: single-step, synchronous, and asynchronous associations. Moreover, we conduct a comprehensive investigation into the MLLMs' zero-shot association capabilities, addressing multiple dimensions, including three distinct memory strategies, both open-source and closed-source MLLMs, cutting-edge Mixture-of-Experts (MoE) models, and the involvement of human experts. Our systematic investigation shows that current open-source MLLMs consistently exhibit poor capability in our association tasks, even the currently state-of-the-art GPT-4V(vision) also has a significant gap compared to humans. We believe our benchmark would pave the way for future MLLM studies. $\textit{Our data and code are available at:}$ https://mvig-rhos.com/llm_inception.
comment: Accepted by ICLR 2025. Project page: https://mvig-rhos.com/llm_inception
♻ ☆ NL2FOL: Translating Natural Language to First-Order Logic for Logical Fallacy Detection
Translating natural language into formal language such as First-Order Logic (FOL) is a foundational challenge in NLP with wide-ranging applications in automated reasoning, misinformation tracking, and knowledge validation. In this paper, we introduce Natural Language to First-Order Logic (NL2FOL), a framework to autoformalize natural language to FOL step by step using Large Language Models (LLMs). Our approach addresses key challenges in this translation process, including the integration of implicit background knowledge. By leveraging structured representations generated by NL2FOL, we use Satisfiability Modulo Theory (SMT) solvers to reason about the logical validity of natural language statements. We present logical fallacy detection as a case study to evaluate the efficacy of NL2FOL. Being neurosymbolic, our approach also provides interpretable insights into the reasoning process and demonstrates robustness without requiring model fine-tuning or labeled training data. Our framework achieves strong performance on multiple datasets. On the LOGIC dataset, NL2FOL achieves an F1-score of 78%, while generalizing effectively to the LOGICCLIMATE dataset with an F1-score of 80%.
♻ ☆ Identifying Drift, Diffusion, and Causal Structure from Temporal Snapshots
Stochastic differential equations (SDEs) are a fundamental tool for modelling dynamic processes, including gene regulatory networks (GRNs), contaminant transport, financial markets, and image generation. However, learning the underlying SDE from data is a challenging task, especially if individual trajectories are not observable. Motivated by burgeoning research in single-cell datasets, we present the first comprehensive approach for jointly identifying the drift and diffusion of an SDE from its temporal marginals. Assuming linear drift and additive diffusion, we prove that these parameters are identifiable from marginals if and only if the initial distribution lacks any generalized rotational symmetries. We further prove that the causal graph of any SDE with additive diffusion can be recovered from the SDE parameters. To complement this theory, we adapt entropy-regularized optimal transport to handle anisotropic diffusion, and introduce APPEX (Alternating Projection Parameter Estimation from $X_0$), an iterative algorithm designed to estimate the drift, diffusion, and causal graph of an additive noise SDE, solely from temporal marginals. We show that APPEX iteratively decreases Kullback-Leibler divergence to the true solution, and demonstrate its effectiveness on simulated data from linear additive noise SDEs.
♻ ☆ Does SGD really happen in tiny subspaces? ICLR 2025
Understanding the training dynamics of deep neural networks is challenging due to their high-dimensional nature and intricate loss landscapes. Recent studies have revealed that, along the training trajectory, the gradient approximately aligns with a low-rank top eigenspace of the training loss Hessian, referred to as the dominant subspace. Given this alignment, this paper explores whether neural networks can be trained within the dominant subspace, which, if feasible, could lead to more efficient training methods. Our primary observation is that when the SGD update is projected onto the dominant subspace, the training loss does not decrease further. This suggests that the observed alignment between the gradient and the dominant subspace is spurious. Surprisingly, projecting out the dominant subspace proves to be just as effective as the original update, despite removing the majority of the original update component. We observe similar behavior across practical setups, including the large learning rate regime (also known as Edge of Stability), Sharpness-Aware Minimization, momentum, and adaptive optimizers. We discuss the main causes and implications of this spurious alignment, shedding light on the dynamics of neural network training.
comment: Published at ICLR 2025
♻ ☆ Asymptotic Behavior of Adversarial Training Estimator under $\ell_\infty$-Perturbation
Adversarial training has been proposed to protect machine learning models against adversarial attacks. This paper focuses on adversarial training under $\ell_\infty$-perturbation, which has recently attracted much research attention. The asymptotic behavior of the adversarial training estimator is investigated in the generalized linear model. The results imply that the asymptotic distribution of the adversarial training estimator under $\ell_\infty$-perturbation could put a positive probability mass at $0$ when the true parameter is $0$, providing a theoretical guarantee of the associated sparsity-recovery ability. Alternatively, a two-step procedure is proposed -- adaptive adversarial training, which could further improve the performance of adversarial training under $\ell_\infty$-perturbation. Specifically, the proposed procedure could achieve asymptotic variable-selection consistency and unbiasedness. Numerical experiments are conducted to show the sparsity-recovery ability of adversarial training under $\ell_\infty$-perturbation and to compare the empirical performance between classic adversarial training and adaptive adversarial training.
♻ ☆ Matryoshka Quantization
Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a trade-off in model quality; int2, in particular, is known to severely degrade model quality. Consequently, practitioners are often forced to maintain multiple models with different quantization levels or serve a single model that best satisfies the quality-latency trade-off. On the other hand, integer data types, such as int8, inherently possess a nested (Matryoshka) structure where smaller bit-width integers, like int4 or int2, are nested within the most significant bits. Leveraging this insight, in this paper, we propose Matryoshka Quantization (MatQuant), a novel multi-scale quantization technique that alleviates the aforementioned challenge. This technique allows us to train and maintain a single quantized model but serve it with the precision demanded by the deployment. Furthermore, leveraging MatQuant's co-training and co-distillation regularization, int2 precision models extracted by MatQuant outperform standard int2 quantization by up to to 4% and 7% with OmniQuant and QAT as base algorithms respectively. Finally, we demonstrate that by using an extra bit to represent outliers, a model with an effective precision of 2.05-bit gives an additional 6% improvement with OmniQuant as the base algorithm.
♻ ☆ InductionBench: LLMs Fail in the Simplest Complexity Class
Large language models (LLMs) have shown remarkable improvements in reasoning and many existing benchmarks have been addressed by models such as o1 and o3 either fully or partially. However, a majority of these benchmarks emphasize deductive reasoning, including mathematical and coding tasks in which rules such as mathematical axioms or programming syntax are clearly defined, based on which LLMs can plan and apply these rules to arrive at a solution. In contrast, inductive reasoning, where one infers the underlying rules from observed data, remains less explored. Such inductive processes lie at the heart of scientific discovery, as they enable researchers to extract general principles from empirical observations. To assess whether LLMs possess this capacity, we introduce InductionBench, a new benchmark designed to evaluate the inductive reasoning ability of LLMs. Our experimental findings reveal that even the most advanced models available struggle to master the simplest complexity classes within the subregular hierarchy of functions, highlighting a notable deficiency in current LLMs' inductive reasoning capabilities. Coda and data are available https://github.com/Wenyueh/inductive_reasoning_benchmark.
comment: 24 pages, 7 figures
♻ ☆ Error Bounds for Physics-Informed Neural Networks in Fokker-Planck PDEs
Stochastic differential equations are commonly used to describe the evolution of stochastic processes. The state uncertainty of such processes is best represented by the probability density function (PDF), whose evolution is governed by the Fokker-Planck partial differential equation (FP-PDE). However, it is generally infeasible to solve the FP-PDE in closed form. In this work, we show that physics-informed neural networks (PINNs) can be trained to approximate the solution PDF. Our main contribution is the analysis of PINN approximation error: we develop a theoretical framework to construct tight error bounds using PINNs. In addition, we derive a practical error bound that can be efficiently constructed with standard training methods. We discuss that this error-bound framework generalizes to approximate solutions of other linear PDEs. Empirical results on nonlinear, high-dimensional, and chaotic systems validate the correctness of our error bounds while demonstrating the scalability of PINNs and their significant computational speedup in obtaining accurate PDF solutions compared to the Monte Carlo approach.
comment: paper under review
♻ ☆ CAMEx: Curvature-aware Merging of Experts ICLR 2025
Existing methods for merging experts during model training and fine-tuning predominantly rely on Euclidean geometry, which assumes a flat parameter space. This assumption can limit the model's generalization ability, especially during the pre-training phase, where the parameter manifold might exhibit more complex curvature. Curvature-aware merging methods typically require additional information and computational resources to approximate the Fisher Information Matrix, adding memory overhead. In this paper, we introduce CAMEx (Curvature-Aware Merging of Experts), a novel expert merging protocol that incorporates natural gradients to account for the non-Euclidean curvature of the parameter manifold. By leveraging natural gradients, CAMEx adapts more effectively to the structure of the parameter space, improving alignment between model updates and the manifold's geometry. This approach enhances both pre-training and fine-tuning, resulting in better optimization trajectories and improved generalization without the substantial memory overhead typically associated with curvature-aware methods. Our contributions are threefold: (1) CAMEx significantly outperforms traditional Euclidean-based expert merging techniques across various natural language processing tasks, leading to enhanced performance during pre-training and fine-tuning; (2) we introduce a dynamic merging architecture that optimizes resource utilization, achieving high performance while reducing computational costs, facilitating efficient scaling of large language models; and (3) we provide both theoretical and empirical evidence to demonstrate the efficiency of our proposed method. The code is publicly available at: https://github.com/kpup1710/CAMEx.
comment: 10 pages, 5 Figures, 7 Tables. Published at ICLR 2025
♻ ☆ Disparate Model Performance and Stability in Machine Learning Clinical Support for Diabetes and Heart Diseases
Machine Learning (ML) algorithms are vital for supporting clinical decision-making in biomedical informatics. However, their predictive performance can vary across demographic groups, often due to the underrepresentation of historically marginalized populations in training datasets. The investigation reveals widespread sex- and age-related inequities in chronic disease datasets and their derived ML models. Thus, a novel analytical framework is introduced, combining systematic arbitrariness with traditional metrics like accuracy and data complexity. The analysis of data from over 25,000 individuals with chronic diseases revealed mild sex-related disparities, favoring predictive accuracy for males, and significant age-related differences, with better accuracy for younger patients. Notably, older patients showed inconsistent predictive accuracy across seven datasets, linked to higher data complexity and lower model performance. This highlights that representativeness in training data alone does not guarantee equitable outcomes, and model arbitrariness must be addressed before deploying models in clinical settings.
comment: This paper will be presented in American Medical Informatics Association (AMIA) Informatics Summit Conference 2025 (Pittsburgh, PA). 10 pages, 2 figures, 5 tables
♻ ☆ "FRAME: Forward Recursive Adaptive Model Extraction-A Technique for Advance Feature Selection"
The challenges in feature selection, particularly in balancing model accuracy, interpretability, and computational efficiency, remain a critical issue in advancing machine learning methodologies. To address these complexities, this study introduces a novel hybrid approach, the Forward Recursive Adaptive Model Extraction Technique (FRAME), which combines Forward Selection and Recursive Feature Elimination (RFE) to enhance feature selection across diverse datasets. By combining the exploratory capabilities of Forward Selection with the refinement strengths of RFE, FRAME systematically identifies optimal feature subsets, striking a harmonious trade-off between experimentation and precision. A comprehensive evaluation of FRAME is conducted against traditional methods such as SelectKBest and Lasso Regression, using high-dimensional, noisy, and heterogeneous datasets. The results demonstrate that FRAME consistently delivers superior predictive performance based on downstream machine learning evaluation metrics. It efficiently performs dimensionality reduction with strong model performance, thus being especially useful for applications that need interpretable and accurate predictions, e.g., biomedical diagnostics. This research emphasizes the need to evaluate feature selection techniques on diverse datasets to test their robustness and generalizability. The results indicate that FRAME has great potential for further development, especially by incorporating deep learning frameworks for adaptive and real-time feature selection in dynamic settings. By advancing feature selection methodologies, FRAME offers a practical and effective solution to improve machine learning applications across multiple domains.
comment: Updated version with refinements before JMLR submission. Improved clarity, expanded literature review, refined methodology, updated experimental results, and enhanced conclusion. FRAME's scalability, deep learning integration, and real-world applications are further highlighted
♻ ☆ IOHunter: Graph Foundation Model to Uncover Online Information Operations AAAI 2025
Social media platforms have become vital spaces for public discourse, serving as modern agor\`as where a wide range of voices influence societal narratives. However, their open nature also makes them vulnerable to exploitation by malicious actors, including state-sponsored entities, who can conduct information operations (IOs) to manipulate public opinion. The spread of misinformation, false news, and misleading claims threatens democratic processes and societal cohesion, making it crucial to develop methods for the timely detection of inauthentic activity to protect the integrity of online discourse. In this work, we introduce a methodology designed to identify users orchestrating information operations, a.k.a. IO drivers, across various influence campaigns. Our framework, named IOHunter, leverages the combined strengths of Language Models and Graph Neural Networks to improve generalization in supervised, scarcely-supervised, and cross-IO contexts. Our approach achieves state-of-the-art performance across multiple sets of IOs originating from six countries, significantly surpassing existing approaches. This research marks a step toward developing Graph Foundation Models specifically tailored for the task of IO detection on social media platforms.
comment: Accepted at AAAI 2025
♻ ☆ Self-Supervised Iterative Refinement for Anomaly Detection in Industrial Quality Control
This study introduces the Iterative Refinement Process (IRP), a robust anomaly detection methodology designed for high-stakes industrial quality control. The IRP enhances defect detection accuracy through a cyclic data refinement strategy, iteratively removing misleading data points to improve model performance and robustness. We validate the IRP's effectiveness using two benchmark datasets, Kolektor SDD2 (KSDD2) and MVTec AD, covering a wide range of industrial products and defect types. Our experimental results demonstrate that the IRP consistently outperforms traditional anomaly detection models, particularly in environments with high noise levels. This study highlights the IRP's potential to significantly enhance anomaly detection processes in industrial settings, effectively managing the challenges of sparse and noisy data.
comment: Accepted to VISAPP 2025
♻ ☆ Gumbel Counterfactual Generation From Language Models ICLR 2025
Understanding and manipulating the causal generation mechanisms in language models is essential for controlling their behavior. Previous work has primarily relied on techniques such as representation surgery -- e.g., model ablations or manipulation of linear subspaces tied to specific concepts -- to \emph{intervene} on these models. To understand the impact of interventions precisely, it is useful to examine \emph{counterfactuals} -- e.g., how a given sentence would have appeared had it been generated by the model following a specific intervention. We highlight that counterfactual reasoning is conceptually distinct from interventions, as articulated in Pearl's causal hierarchy. Based on this observation, we propose a framework for generating true string counterfactuals by reformulating language models as a structural equation model using the Gumbel-max trick, which we called Gumbel counterfactual generation. This reformulation allows us to model the joint distribution over original strings and their counterfactuals resulting from the same instantiation of the sampling noise. We develop an algorithm based on hindsight Gumbel sampling that allows us to infer the latent noise variables and generate counterfactuals of observed strings. Our experiments demonstrate that the approach produces meaningful counterfactuals while at the same time showing that commonly used intervention techniques have considerable undesired side effects.
comment: Accepted in ICLR 2025
♻ ☆ Using High-Level Patterns to Estimate How Humans Predict a Robot will Behave
Humans interacting with robots often form predictions of what the robot will do next. For instance, based on the recent behavior of an autonomous car, a nearby human driver might predict that the car is going to remain in the same lane. It is important for the robot to understand the human's prediction for safe and seamless interaction: e.g., if the autonomous car knows the human thinks it is not merging -- but the autonomous car actually intends to merge -- then the car can adjust its behavior to prevent an accident. Prior works typically assume that humans make precise predictions of robot behavior. However, recent research on human-human prediction suggests the opposite: humans tend to approximate other agents by predicting their high-level behaviors. We apply this finding to develop a second-order theory of mind approach that enables robots to estimate how humans predict they will behave. To extract these high-level predictions directly from data, we embed the recent human and robot trajectories into a discrete latent space. Each element of this latent space captures a different type of behavior (e.g., merging in front of the human, remaining in the same lane) and decodes into a vector field across the state space that is consistent with the underlying behavior type. We hypothesize that our resulting high-level and course predictions of robot behavior will correspond to actual human predictions. We provide initial evidence in support of this hypothesis through proof-of-concept simulations, testing our method's predictions against those of real users, and experiments on a real-world interactive driving dataset.
♻ ☆ Multi-Modal and Multi-Attribute Generation of Single Cells with CFGen
Generative modeling of single-cell RNA-seq data is crucial for tasks like trajectory inference, batch effect removal, and simulation of realistic cellular data. However, recent deep generative models simulating synthetic single cells from noise operate on pre-processed continuous gene expression approximations, overlooking the discrete nature of single-cell data, which limits their effectiveness and hinders the incorporation of robust noise models. Additionally, aspects like controllable multi-modal and multi-label generation of cellular data remain underexplored. This work introduces CellFlow for Generation (CFGen), a flow-based conditional generative model that preserves the inherent discreteness of single-cell data. CFGen generates whole-genome multi-modal single-cell data reliably, improving the recovery of crucial biological data characteristics while tackling relevant generative tasks such as rare cell type augmentation and batch correction. We also introduce a novel framework for compositional data generation using Flow Matching. By showcasing CFGen on a diverse set of biological datasets and settings, we provide evidence of its value to the fields of computational biology and deep generative models.
comment: 41 pages, 22 figures
♻ ☆ Enhancing Fairness in Unsupervised Graph Anomaly Detection through Disentanglement
Graph anomaly detection (GAD) is increasingly crucial in various applications, ranging from financial fraud detection to fake news detection. However, current GAD methods largely overlook the fairness problem, which might result in discriminatory decisions skewed toward certain demographic groups defined on sensitive attributes (e.g., gender, religion, ethnicity, etc.). This greatly limits the applicability of these methods in real-world scenarios in light of societal and ethical restrictions. To address this critical gap, we make the first attempt to integrate fairness with utility in GAD decision-making. Specifically, we devise a novel DisEntangle-based FairnEss-aware aNomaly Detection framework on the attributed graph, named DEFEND. DEFEND first introduces disentanglement in GNNs to capture informative yet sensitive-irrelevant node representations, effectively reducing societal bias inherent in graph representation learning. Besides, to alleviate discriminatory bias in evaluating anomalous nodes, DEFEND adopts a reconstruction-based anomaly detection, which concentrates solely on node attributes without incorporating any graph structure. Additionally, given the inherent association between input and sensitive attributes, DEFEND constrains the correlation between the reconstruction error and the predicted sensitive attributes. Our empirical evaluations on real-world datasets reveal that DEFEND performs effectively in GAD and significantly enhances fairness compared to state-of-the-art baselines. To foster reproducibility, our code is available at https://github.com/AhaChang/DEFEND.
comment: Accepted to TMLR. Code available at https://github.com/AhaChang/DEFEND
♻ ☆ Towards a General Time Series Anomaly Detector with Adaptive Bottlenecks and Dual Adversarial Decoders ICLR 2025
Time series anomaly detection plays a vital role in a wide range of applications. Existing methods require training one specific model for each dataset, which exhibits limited generalization capability across different target datasets, hindering anomaly detection performance in various scenarios with scarce training data. Aiming at this problem, we propose constructing a general time series anomaly detection model, which is pre-trained on extensive multi-domain datasets and can subsequently apply to a multitude of downstream scenarios. The significant divergence of time series data across different domains presents two primary challenges in building such a general model: (1) meeting the diverse requirements of appropriate information bottlenecks tailored to different datasets in one unified model, and (2) enabling distinguishment between multiple normal and abnormal patterns, both are crucial for effective anomaly detection in various target scenarios. To tackle these two challenges, we propose a General time series anomaly Detector with Adaptive Bottlenecks and Dual Adversarial Decoders (DADA), which enables flexible selection of bottlenecks based on different data and explicitly enhances clear differentiation between normal and abnormal series. We conduct extensive experiments on nine target datasets from different domains. After pre-training on multi-domain data, DADA, serving as a zero-shot anomaly detector for these datasets, still achieves competitive or even superior results compared to those models tailored to each specific dataset. The code is made available at https://github.com/decisionintelligence/DADA.
comment: Accepted by the 13th International Conference on Learning Representations (ICLR 2025)
♻ ☆ Efficient Learning With Sine-Activated Low-rank Matrices ICLR 2025
Low-rank decomposition has emerged as a vital tool for enhancing parameter efficiency in neural network architectures, gaining traction across diverse applications in machine learning. These techniques significantly lower the number of parameters, striking a balance between compactness and performance. However, a common challenge has been the compromise between parameter efficiency and the accuracy of the model, where reduced parameters often lead to diminished accuracy compared to their full-rank counterparts. In this work, we propose a novel theoretical framework that integrates a sinusoidal function within the low-rank decomposition process. This approach not only preserves the benefits of the parameter efficiency characteristic of low-rank methods but also increases the decomposition's rank, thereby enhancing model performance. Our method proves to be a plug in enhancement for existing low-rank models, as evidenced by its successful application in Vision Transformers (ViT), Large Language Models (LLMs), Neural Radiance Fields (NeRF) and 3D shape modelling.
comment: The first two authors contributed equally. Paper accepted at ICLR 2025
♻ ☆ A Decade's Battle on Dataset Bias: Are We There Yet? ICLR 2025
We revisit the "dataset classification" experiment suggested by Torralba & Efros (2011) a decade ago, in the new era with large-scale, diverse, and hopefully less biased datasets as well as more capable neural network architectures. Surprisingly, we observe that modern neural networks can achieve excellent accuracy in classifying which dataset an image is from: e.g., we report 84.7% accuracy on held-out validation data for the three-way classification problem consisting of the YFCC, CC, and DataComp datasets. Our further experiments show that such a dataset classifier could learn semantic features that are generalizable and transferable, which cannot be explained by memorization. We hope our discovery will inspire the community to rethink issues involving dataset bias.
comment: Published in ICLR 2025 (Oral Presentation)
♻ ☆ Iterated $Q$-Network: Beyond One-Step Bellman Updates in Deep Reinforcement Learning
The vast majority of Reinforcement Learning methods is largely impacted by the computation effort and data requirements needed to obtain effective estimates of action-value functions, which in turn determine the quality of the overall performance and the sample-efficiency of the learning procedure. Typically, action-value functions are estimated through an iterative scheme that alternates the application of an empirical approximation of the Bellman operator and a subsequent projection step onto a considered function space. It has been observed that this scheme can be potentially generalized to carry out multiple iterations of the Bellman operator at once, benefiting the underlying learning algorithm. However, till now, it has been challenging to effectively implement this idea, especially in high-dimensional problems. In this paper, we introduce iterated $Q$-Network (i-QN), a novel principled approach that enables multiple consecutive Bellman updates by learning a tailored sequence of action-value functions where each serves as the target for the next. We show that i-QN is theoretically grounded and that it can be seamlessly used in value-based and actor-critic methods. We empirically demonstrate the advantages of i-QN in Atari $2600$ games and MuJoCo continuous control problems.
comment: Published at TMLR: https://openreview.net/forum?id=Lt2H8Bd8jF
♻ ☆ AI, Meet Human: Learning Paradigms for Hybrid Decision Making Systems
Everyday we increasingly rely on machine learning models to automate and support high-stake tasks and decisions. This growing presence means that humans are now constantly interacting with machine learning-based systems, training and using models everyday. Several different techniques in computer science literature account for the human interaction with machine learning systems, but their classification is sparse and the goals varied. This survey proposes a taxonomy of Hybrid Decision Making Systems, providing both a conceptual and technical framework for understanding how current computer science literature models interaction between humans and machines.
♻ ☆ Signature Kernel Conditional Independence Tests in Causal Discovery for Stochastic Processes
Inferring the causal structure underlying stochastic dynamical systems from observational data holds great promise in domains ranging from science and health to finance. Such processes can often be accurately modeled via stochastic differential equations (SDEs), which naturally imply causal relationships via "which variables enter the differential of which other variables". In this paper, we develop conditional independence (CI) constraints on coordinate processes over selected intervals that are Markov with respect to the acyclic dependence graph (allowing self-loops) induced by a general SDE model. We then provide a sound and complete causal discovery algorithm, capable of handling both fully and partially observed data, and uniquely recovering the underlying or induced ancestral graph by exploiting time directionality assuming a CI oracle. Finally, to make our algorithm practically usable, we also propose a flexible, consistent signature kernel-based CI test to infer these constraints from data. We extensively benchmark the CI test in isolation and as part of our causal discovery algorithms, outperforming existing approaches in SDE models and beyond.
♻ ☆ Geometry-Aware Approaches for Balancing Performance and Theoretical Guarantees in Linear Bandits
This paper is motivated by recent research in the $d$-dimensional stochastic linear bandit literature, which has revealed an unsettling discrepancy: algorithms like Thompson sampling and Greedy demonstrate promising empirical performance, yet this contrasts with their pessimistic theoretical regret bounds. The challenge arises from the fact that while these algorithms may perform poorly in certain problem instances, they generally excel in typical instances. To address this, we propose a new data-driven technique that tracks the geometric properties of the uncertainty ellipsoid around the main problem parameter. This methodology enables us to formulate a data-driven frequentist regret bound, which incorporates the geometric information, for a broad class of base algorithms, including Greedy, OFUL, and Thompson sampling. This result allows us to identify and ``course-correct" problem instances in which the base algorithms perform poorly. The course-corrected algorithms achieve the minimax optimal regret of order $\tilde{\mathcal{O}}(d\sqrt{T})$ for a $T$-period decision-making scenario, effectively maintaining the desirable attributes of the base algorithms, including their empirical efficacy. We present simulation results to validate our findings using synthetic and real data.
Computation and Language 49
♻ ☆ ConFit v2: Improving Resume-Job Matching using Hypothetical Resume Embedding and Runner-Up Hard-Negative Mining
A reliable resume-job matching system helps a company recommend suitable candidates from a pool of resumes and helps a job seeker find relevant jobs from a list of job posts. However, since job seekers apply only to a few jobs, interaction labels in resume-job datasets are sparse. We introduce ConFit v2, an improvement over ConFit to tackle this sparsity problem. We propose two techniques to enhance the encoder's contrastive training process: augmenting job data with hypothetical reference resume generated by a large language model; and creating high-quality hard negatives from unlabeled resume/job pairs using a novel hard-negative mining strategy. We evaluate ConFit v2 on two real-world datasets and demonstrate that it outperforms ConFit and prior methods (including BM25 and OpenAI text-embedding-003), achieving an average absolute improvement of 13.8% in recall and 17.5% in nDCG across job-ranking and resume-ranking tasks.
comment: arXiv admin note: text overlap with arXiv:2401.16349
♻ ☆ Inference to the Best Explanation in Large Language Models
While Large Language Models (LLMs) have found success in real-world applications, their underlying explanatory process is still poorly understood. This paper proposes IBE-Eval, a framework inspired by philosophical accounts on Inference to the Best Explanation (IBE) to advance the interpretation and evaluation of LLMs' explanations. IBE-Eval estimates the plausibility of natural language explanations through a combination of explicit logical and linguistic features including: consistency, parsimony, coherence, and uncertainty. Extensive experiments are conducted on Causal Question Answering (CQA), where \textit{IBE-Eval} is tasked to select the most plausible causal explanation amongst competing ones generated by LLMs (i.e., GPT 3.5 and Llama 2). The experiments reveal that IBE-Eval can successfully identify the best explanation with up to 77\% accuracy ($\approx 27\%$ above random), improving upon a GPT 3.5-as-a-Judge baseline ($\approx+17\%$) while being intrinsically more efficient and interpretable. Additional analyses suggest that, despite model-specific variances, LLM-generated explanations tend to conform to IBE criteria and that IBE-Eval is significantly correlated with human judgment, opening up opportunities for future development of automated explanation verification tools.
♻ ☆ Inference Scaling for Long-Context Retrieval Augmented Generation ICLR 2025
The scaling of inference computation has unlocked the potential of long-context large language models (LLMs) across diverse settings. For knowledge-intensive tasks, the increased compute is often allocated to incorporate more external knowledge. However, without effectively utilizing such knowledge, solely expanding context does not always enhance performance. In this work, we investigate inference scaling for retrieval augmented generation (RAG), exploring the combination of multiple strategies beyond simply increasing the quantity of knowledge, including in-context learning and iterative prompting. These strategies provide additional flexibility to scale test-time computation (e.g., by increasing retrieved documents or generation steps), thereby enhancing LLMs' ability to effectively acquire and utilize contextual information. We address two key questions: (1) How does RAG performance benefit from the scaling of inference computation when optimally configured? (2) Can we predict the optimal test-time compute allocation for a given budget by modeling the relationship between RAG performance and inference parameters? Our observations reveal that increasing inference computation leads to nearly linear gains in RAG performance when optimally allocated, a relationship we describe as the inference scaling laws for RAG. Building on this, we further develop the computation allocation model to estimate RAG performance across different inference configurations. The model predicts optimal inference parameters under various computation constraints, which align closely with the experimental results. By applying these optimal configurations, we demonstrate that scaling inference compute on long-context LLMs achieves up to 58.9% gains on benchmark datasets compared to standard RAG.
comment: ICLR 2025
♻ ☆ Distributed Speculative Inference (DSI): Speculation Parallelism for Provably Faster Lossless Language Model Inference ICLR 2025
This paper introduces distributed speculative inference (DSI), a novel inference algorithm that is provably faster than speculative inference (SI) [leviathan2023, chen2023, miao2024, sun2025, timor2025] and standard autoregressive inference (non-SI). Like other SI algorithms, DSI operates on frozen language models (LMs), requiring no training or architectural modifications, and it preserves the target distribution. Prior studies on SI have demonstrated empirical speedups over non-SI--but rely on sufficiently fast and accurate drafters, which are often unavailable in practice. We identify a gap where SI can be slower than non-SI if drafters are too slow or inaccurate. We close this gap by proving that DSI is faster than both SI and non-SI--given any drafters. DSI is therefore not only faster than SI, but also unlocks the acceleration of LMs for which SI fails. DSI leverages speculation parallelism (SP), a novel type of task parallelism, to orchestrate target and drafter instances that overlap in time, establishing a new foundational tradeoff between computational resources and latency. Our simulations show that DSI is 1.29-1.92x faster than SI in single-node setups for various off-the-shelf LMs and tasks. We open-source all our code.
comment: Published at ICLR 2025. (Link: https://openreview.net/forum?id=cJd1BgZ9CS)
♻ ☆ Prompting Fairness: Integrating Causality to Debias Large Language Models
Large language models (LLMs), despite their remarkable capabilities, are susceptible to generating biased and discriminatory responses. As LLMs increasingly influence high-stakes decision-making (e.g., hiring and healthcare), mitigating these biases becomes critical. In this work, we propose a causality-guided debiasing framework to tackle social biases, aiming to reduce the objectionable dependence between LLMs' decisions and the social information in the input. Our framework introduces a novel perspective to identify how social information can affect an LLM's decision through different causal pathways. Leveraging these causal insights, we outline principled prompting strategies that regulate these pathways through selection mechanisms. This framework not only unifies existing prompting-based debiasing techniques, but also opens up new directions for reducing bias by encouraging the model to prioritize fact-based reasoning over reliance on biased social cues. We validate our framework through extensive experiments on real-world datasets across multiple domains, demonstrating its effectiveness in debiasing LLM decisions, even with only black-box access to the model.
comment: 24 pages, 10 figures
♻ ☆ Leveraging Dual Process Theory in Language Agent Framework for Real-time Simultaneous Human-AI Collaboration
Agents built on large language models (LLMs) have excelled in turn-by-turn human-AI collaboration but struggle with simultaneous tasks requiring real-time interaction. Latency issues and the challenge of inferring variable human strategies hinder their ability to make autonomous decisions without explicit instructions. Through experiments with current independent System 1 and System 2 methods, we validate the necessity of using Dual Process Theory (DPT) in real-time tasks. We propose DPT-Agent, a novel language agent framework that integrates System 1 and System 2 for efficient real-time simultaneous human-AI collaboration. DPT-Agent's System 1 uses a Finite-state Machine (FSM) and code-as-policy for fast, intuitive, and controllable decision-making. DPT-Agent's System 2 integrates Theory of Mind (ToM) and asynchronous reflection to infer human intentions and perform reasoning-based autonomous decisions. We demonstrate the effectiveness of DPT-Agent through further experiments with rule-based agents and human collaborators, showing significant improvements over mainstream LLM-based frameworks. DPT-Agent can effectively help LLMs convert correct slow thinking and reasoning into executable actions, thereby improving performance. To the best of our knowledge, DPT-Agent is the first language agent framework that achieves successful real-time simultaneous human-AI collaboration autonomously. Code of DPT-Agent can be found in https://github.com/sjtu-marl/DPT-Agent.
comment: Preprint under review. Update the experimental results of the DeepSeek-R1 series models, o3-mini-high and o3-mini-medium
♻ ☆ Harnessing Multiple Large Language Models: A Survey on LLM Ensemble
LLM Ensemble -- which involves the comprehensive use of multiple large language models (LLMs), each aimed at handling user queries during downstream inference, to benefit from their individual strengths -- has gained substantial attention recently. The widespread availability of LLMs, coupled with their varying strengths and out-of-the-box usability, has profoundly advanced the field of LLM Ensemble. This paper presents the first systematic review of recent developments in LLM Ensemble. First, we introduce our taxonomy of LLM Ensemble and discuss several related research problems. Then, we provide a more in-depth classification of the methods under the broad categories of "ensemble-before-inference, ensemble-during-inference, ensemble-after-inference'', and review all relevant methods. Finally, we introduce related benchmarks and applications, summarize existing studies, and suggest several future research directions. A curated list of papers on LLM Ensemble is available at https://github.com/junchenzhi/Awesome-LLM-Ensemble.
comment: 9 pages, 2 figures, codebase: https://github.com/junchenzhi/Awesome-LLM-Ensemble
♻ ☆ Large Language Models Meet Symbolic Provers for Logical Reasoning Evaluation ICLR 2025
First-order logic (FOL) reasoning, which involves sequential deduction, is pivotal for intelligent systems and serves as a valuable task for evaluating reasoning capabilities, particularly in chain-of-thought (CoT) contexts. Existing benchmarks often rely on extensive human annotation or handcrafted templates, making it difficult to achieve the necessary complexity, scalability, and diversity for robust evaluation. To address these limitations, we propose a novel framework called ProverGen that synergizes the generative strengths of Large Language Models (LLMs) with the rigor and precision of symbolic provers, enabling the creation of a scalable, diverse, and high-quality FOL reasoning dataset, ProverQA. ProverQA is also distinguished by its inclusion of accessible and logically coherent intermediate reasoning steps for each problem. Our evaluation shows that state-of-the-art LLMs struggle to solve ProverQA problems, even with CoT prompting, highlighting the dataset's challenging nature. We also finetune Llama3.1-8B-Instruct on a separate training set generated by our framework. The finetuned model demonstrates consistent improvements on both in-distribution and out-of-distribution test sets, suggesting the value of our proposed data generation framework. Code available at: https://github.com/opendatalab/ProverGen
comment: Accepted by ICLR 2025
♻ ☆ Generative causal testing to bridge data-driven models and scientific theories in language neuroscience
Representations from large language models are highly effective at predicting BOLD fMRI responses to language stimuli. However, these representations are largely opaque: it is unclear what features of the language stimulus drive the response in each brain area. We present generative causal testing (GCT), a framework for generating concise explanations of language selectivity in the brain from predictive models and then testing those explanations in follow-up experiments using LLM-generated stimuli.This approach is successful at explaining selectivity both in individual voxels and cortical regions of interest (ROIs), including newly identified microROIs in prefrontal cortex. We show that explanatory accuracy is closely related to the predictive power and stability of the underlying predictive models. Finally, we show that GCT can dissect fine-grained differences between brain areas with similar functional selectivity. These results demonstrate that LLMs can be used to bridge the widening gap between data-driven models and formal scientific theories.
♻ ☆ The Mighty ToRR: A Benchmark for Table Reasoning and Robustness
Despite its real-world significance, model performance on tabular data remains underexplored, leaving uncertainty about which model to rely on and which prompt configuration to adopt. To address this gap, we create ToRR, a benchmark for Table Reasoning and Robustness, measuring model performance and robustness on table-related tasks. The benchmark includes 10 datasets that cover different types of table reasoning capabilities across varied domains. ToRR goes beyond model performance rankings, and is designed to reflect whether models can handle tabular data consistently and robustly, across a variety of common table representation formats. We present a leaderboard as well as comprehensive analyses of the results of leading models over ToRR. Our results reveal a striking pattern of brittle model behavior, where even strong models are unable to perform robustly on tabular data tasks. Although no specific table format leads to consistently better performance, we show that testing over multiple formats is crucial for reliably estimating model capabilities. Moreover, we show that the reliability boost from testing multiple prompts can be equivalent to adding more test examples. Overall, our findings show that table understanding and reasoning tasks remain a significant challenge.
♻ ☆ LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Token ICLR 2025
The advent of real-time large multimodal models (LMMs) like GPT-4o has sparked considerable interest in efficient LMMs. LMM frameworks typically encode visual inputs into vision tokens (continuous representations) and integrate them and textual instructions into the context of large language models (LLMs), where large-scale parameters and numerous context tokens (predominantly vision tokens) result in substantial computational overhead. Previous efforts towards efficient LMMs always focus on replacing the LLM backbone with smaller models, while neglecting the crucial issue of token quantity. In this paper, we introduce LLaVA-Mini, an efficient LMM with minimal vision tokens. To achieve a high compression ratio of vision tokens while preserving visual information, we first analyze how LMMs understand vision tokens and find that most vision tokens only play a crucial role in the early layers of LLM backbone, where they mainly fuse visual information into text tokens. Building on this finding, LLaVA-Mini introduces modality pre-fusion to fuse visual information into text tokens in advance, thereby facilitating the extreme compression of vision tokens fed to LLM backbone into one token. LLaVA-Mini is a unified large multimodal model that can support the understanding of images, high-resolution images, and videos in an efficient manner. Experiments across 11 image-based and 7 video-based benchmarks demonstrate that LLaVA-Mini outperforms LLaVA-v1.5 with just 1 vision token instead of 576. Efficiency analyses reveal that LLaVA-Mini can reduce FLOPs by 77%, deliver low-latency responses within 40 milliseconds, and process over 10,000 frames of video on the GPU hardware with 24GB of memory.
comment: Accepted to ICLR 2025. Code: https://github.com/ictnlp/LLaVA-Mini Model: https://huggingface.co/ICTNLP/llava-mini-llama-3.1-8b
♻ ☆ Steering Large Language Models between Code Execution and Textual Reasoning
While a lot of recent research focuses on enhancing the textual reasoning capabilities of Large Language Models (LLMs) by optimizing the multi-agent framework or reasoning chains, several benchmark tasks can be solved with 100\% success through direct coding, which is more scalable and avoids the computational overhead associated with textual iterating and searching. Textual reasoning has inherent limitations in solving tasks with challenges in math, logics, optimization, and searching, which is unlikely to be solved by simply scaling up the model and data size. The recently released OpenAI GPT Code Interpreter and multi-agent frameworks such as AutoGen have demonstrated remarkable proficiency of integrating code generation and execution to solve complex tasks using LLMs. However, based on our experiments on 7 existing popular methods for steering code/text generation in both single- and multi-turn settings with 14 tasks and 6 types of LLMs (including the new O1-preview), currently there is no optimal method to correctly steer LLMs to write code when needed. We discover some interesting patterns on when models use code vs. textual reasoning with the evolution to task complexity and model sizes, which even result in an astonishingly inverse scaling behavior. We also discover that results from LLM written code are not always better than using textual reasoning, even if the task could be solved through code. To mitigate the above issues, we propose three methods to better steer LLM code/text generation and achieve a notable improvement. The costs of token lengths and runtime are thoroughly discussed for all the methods. We believe the problem of steering LLM code/text generation is critical for future research and has much space for further improvement. Project Page, Datasets, and Codes are available at https://yongchao98.github.io/CodeSteer/.
comment: 32 pages, 12 figures, 12 tables
♻ ☆ When Attention Sink Emerges in Language Models: An Empirical View ICLR 2025
Language Models (LMs) assign significant attention to the first token, even if it is not semantically important, which is known as attention sink. This phenomenon has been widely adopted in applications such as streaming/long context generation, KV cache optimization, inference acceleration, model quantization, and others. Despite its widespread use, a deep understanding of attention sink in LMs is still lacking. In this work, we first demonstrate that attention sinks exist universally in LMs with various inputs, even in small models. Furthermore, attention sink is observed to emerge during the LM pre-training, motivating us to investigate how optimization, data distribution, loss function, and model architecture in LM pre-training influence its emergence. We highlight that attention sink emerges after effective optimization on sufficient training data. The sink position is highly correlated with the loss function and data distribution. Most importantly, we find that attention sink acts more like key biases, storing extra attention scores, which could be non-informative and not contribute to the value computation. We also observe that this phenomenon (at least partially) stems from tokens' inner dependence on attention scores as a result of softmax normalization. After relaxing such dependence by replacing softmax attention with other attention operations, such as sigmoid attention without normalization, attention sinks do not emerge in LMs up to 1B parameters. The code is available at https://github.com/sail-sg/Attention-Sink.
comment: ICLR 2025 (Spotlight)
♻ ☆ Generating Visual Stories with Grounded and Coreferent Characters
Characters are important in narratives. They move the plot forward, create emotional connections, and embody the story's themes. Visual storytelling methods focus more on the plot and events relating to it, without building the narrative around specific characters. As a result, the generated stories feel generic, with character mentions being absent, vague, or incorrect. To mitigate these issues, we introduce the new task of character-centric story generation and present the first model capable of predicting visual stories with consistently grounded and coreferent character mentions. Our model is finetuned on a new dataset which we build on top of the widely used VIST benchmark. Specifically, we develop an automated pipeline to enrich VIST with visual and textual character coreference chains. We also propose new evaluation metrics to measure the richness of characters and coreference in stories. Experimental results show that our model generates stories with recurring characters which are consistent and coreferent to larger extent compared to baselines and state-of-the-art systems.
♻ ☆ Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates ICLR 2025
Automatic LLM benchmarks, such as AlpacaEval 2.0, Arena-Hard-Auto, and MT-Bench, have become popular for evaluating language models due to their cost-effectiveness and scalability compared to human evaluation. Achieving high win rates on these benchmarks can significantly boost the promotional impact of newly released language models. This promotional benefit may motivate tricks, such as manipulating model output length or style to game win rates, even though several mechanisms have been developed to control length and disentangle style to reduce gameability. Nonetheless, we show that even a "null model" that always outputs a constant response (irrelevant to input instructions) can cheat automatic benchmarks and achieve top-ranked win rates: an 86.5% LC win rate on AlpacaEval 2.0; an 83.0 score on Arena-Hard-Auto; and a 9.55 score on MT-Bench. Moreover, the crafted cheating outputs are transferable because we assume that the instructions of these benchmarks (e.g., 805 samples of AlpacaEval 2.0) are private and cannot be accessed. While our experiments are primarily proof-of-concept, an adversary could use LLMs to generate more imperceptible cheating responses, unethically benefiting from high win rates and promotional impact. Our findings call for the development of anti-cheating mechanisms for reliable automatic benchmarks. The code is available at https://github.com/sail-sg/Cheating-LLM-Benchmarks.
comment: ICLR 2025 (Oral)
♻ ☆ How Does A Text Preprocessing Pipeline Affect Ontology Syntactic Matching?
The generic text preprocessing pipeline, comprising Tokenisation, Normalisation, Stop Words Removal, and Stemming/Lemmatisation, has been implemented in many systems for syntactic ontology matching (OM). However, the lack of standardisation in text preprocessing creates diversity in mapping results. In this paper, we investigate the effect of the text preprocessing pipeline on syntactic OM in 8 Ontology Alignment Evaluation Initiative (OAEI) tracks with 49 distinct alignments. We find that Phase 1 text preprocessing (Tokenisation and Normalisation) is currently more effective than Phase 2 text preprocessing (Stop Words Removal and Stemming/Lemmatisation). To repair the less effective Phase 2 text preprocessing caused by unwanted false mappings, we propose a novel context-based pipeline repair approach that employs an ad hoc check to find common words that cause false mappings. These words are stored in a reserved word set and applied in text preprocessing. The experimental results show that our approach improves the matching correctness and the overall matching performance. We also discuss the integration of the classical text preprocessing pipeline with modern large language models (LLMs). We recommend that LLMs inject the text preprocessing pipeline via function calling to avoid the tendency towards unstable true mappings produced by prompt-based LLM approaches, and use LLMs to repair false mappings generated by the text preprocessing pipeline.
comment: 13 pages, 11 figures, 4 tables
♻ ☆ Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations ICLR 2025
To gain insight into the mechanisms behind machine learning methods, it is crucial to establish connections among the features describing data points. However, these correlations often exhibit a high-dimensional and strongly nonlinear nature, which makes them challenging to detect using standard methods. This paper exploits the entanglement between intrinsic dimensionality and correlation to propose a metric that quantifies the (potentially nonlinear) correlation between high-dimensional manifolds. We first validate our method on synthetic data in controlled environments, showcasing its advantages and drawbacks compared to existing techniques. Subsequently, we extend our analysis to large-scale applications in neural network representations. Specifically, we focus on latent representations of multimodal data, uncovering clear correlations between paired visual and textual embeddings, whereas existing methods struggle significantly in detecting similarity. Our results indicate the presence of highly nonlinear correlation patterns between latent manifolds.
comment: Accepted at ICLR 2025
♻ ☆ Boosting Jailbreak Attack with Momentum ICASSP 2025
Large Language Models (LLMs) have achieved remarkable success across diverse tasks, yet they remain vulnerable to adversarial attacks, notably the well-known jailbreak attack. In particular, the Greedy Coordinate Gradient (GCG) attack has demonstrated efficacy in exploiting this vulnerability by optimizing adversarial prompts through a combination of gradient heuristics and greedy search. However, the efficiency of this attack has become a bottleneck in the attacking process. To mitigate this limitation, in this paper we rethink the generation of the adversarial prompts through an optimization lens, aiming to stabilize the optimization process and harness more heuristic insights from previous optimization iterations. Specifically, we propose the \textbf{M}omentum \textbf{A}ccelerated G\textbf{C}G (\textbf{MAC}) attack, which integrates a momentum term into the gradient heuristic to boost and stabilize the random search for tokens in adversarial prompts. Experimental results showcase the notable enhancement achieved by MAC over baselines in terms of attack success rate and optimization efficiency. Moreover, we demonstrate that MAC can still exhibit superior performance for transfer attacks and models under defense mechanisms. Our code is available at https://github.com/weizeming/momentum-attack-llm.
comment: Accepted by ICASSP 2025
♻ ☆ Developing a Multilingual Dataset and Evaluation Metrics for Code-Switching: A Focus on Hong Kong's Polylingual Dynamics
The existing audio datasets are predominantly tailored towards single languages, overlooking the complex linguistic behaviors of multilingual communities that engage in code-switching. This practice, where individuals frequently mix two or more languages in their daily interactions, is particularly prevalent in multilingual regions such as Hong Kong, China. To bridge this gap, we have developed a 34.8-hour dataset of Mixed Cantonese and English (MCE) audio using our Multi-Agent Data Generation Framework (MADGF). We fine-tuned the open-source multilingual Automatic Speech Recognition (ASR) model, Whisper, with the MCE dataset, leading to impressive zero-shot performance. The traditional metrics overlook important factors such as latency in real-world applications and code-switching scenarios. We have introduced a novel evaluation metric called Fidelity to the Original Audio, Accuracy, and Latency (FAL). This metric aims to overcome the limitations of traditional metrics used to assess ASR systems.
♻ ☆ Derailer-Rerailer: Adaptive Verification for Efficient and Reliable Language Model Reasoning
Large Language Models (LLMs) have shown impressive reasoning capabilities, yet existing prompting methods face a critical trade-off: simple approaches often struggle with complex tasks and reasoning stability, while more sophisticated methods require multiple inferences and substantial computational resources, limiting their practical deployment. To address this challenge, we propose Derailer-Rerailer, a novel framework that adaptively balances reasoning accuracy and computational efficiency. At its core, our framework employs a lightweight Derailer mechanism to assess reasoning stability and selectively triggers an advanced Rerailer verification process only when necessary, thereby optimizing computational resource usage. Extensive evaluation across both open and closed-source models on more than 20 categories of mathematical, symbolic, and commonsense reasoning tasks demonstrates our framework's effectiveness: Derailer-Rerailer achieves significant accuracy improvements (8-11\% across various reasoning tasks) while maintaining 2-3 times better efficiency than existing verification methods, with particularly strong performance in mathematical and symbolic reasoning, offering a practical solution for enhancing LLM reasoning reliability while significantly reducing computational overhead.
♻ ☆ Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models
Representation Engineering (RepE) is a novel paradigm for controlling the behavior of LLMs. Unlike traditional approaches that modify inputs or fine-tune the model, RepE directly manipulates the model's internal representations. As a result, it may offer more effective, interpretable, data-efficient, and flexible control over models' behavior. We present the first comprehensive survey of RepE for LLMs, reviewing the rapidly growing literature to address key questions: What RepE methods exist and how do they differ? For what concepts and problems has RepE been applied? What are the strengths and weaknesses of RepE compared to other methods? To answer these, we propose a unified framework describing RepE as a pipeline comprising representation identification, operationalization, and control. We posit that while RepE methods offer significant potential, challenges remain, including managing multiple concepts, ensuring reliability, and preserving models' performance. Towards improving RepE, we identify opportunities for experimental and methodological improvements and construct a guide for best practices.
♻ ☆ DiscoGraMS: Enhancing Movie Screen-Play Summarization using Movie Character-Aware Discourse Graph NAACL 2025
Summarizing movie screenplays presents a unique set of challenges compared to standard document summarization. Screenplays are not only lengthy, but also feature a complex interplay of characters, dialogues, and scenes, with numerous direct and subtle relationships and contextual nuances that are difficult for machine learning models to accurately capture and comprehend. Recent attempts at screenplay summarization focus on fine-tuning transformer-based pre-trained models, but these models often fall short in capturing long-term dependencies and latent relationships, and frequently encounter the "lost in the middle" issue. To address these challenges, we introduce DiscoGraMS, a novel resource that represents movie scripts as a movie character-aware discourse graph (CaD Graph). This approach is well-suited for various downstream tasks, such as summarization, question-answering, and salience detection. The model aims to preserve all salient information, offering a more comprehensive and faithful representation of the screenplay's content. We further explore a baseline method that combines the CaD Graph with the corresponding movie script through a late fusion of graph and text modalities, and we present very initial promising results.
comment: Accepted at NAACL 2025 (Main)
♻ ☆ Market-Derived Financial Sentiment Analysis: Context-Aware Language Models for Crypto Forecasting
Financial Sentiment Analysis (FSA) traditionally relies on human-annotated sentiment labels to infer investor sentiment and forecast market movements. However, inferring the potential market impact of words based on their human-perceived intentions is inherently challenging. We hypothesize that the historical market reactions to words, offer a more reliable indicator of their potential impact on markets than subjective sentiment interpretations by human annotators. To test this hypothesis, a market-derived labeling approach is proposed to assign tweet labels based on ensuing short-term price trends, enabling the language model to capture the relationship between textual signals and market dynamics directly. A domain-specific language model was fine-tuned on these labels, achieving up to an 11% improvement in short-term trend prediction accuracy over traditional sentiment-based benchmarks. Moreover, by incorporating market and temporal context through prompt-tuning, the proposed context-aware language model demonstrated an accuracy of 89.6% on a curated dataset of 227 impactful Bitcoin-related news events with significant market impacts. Aggregating daily tweet predictions into trading signals, our method outperformed traditional fusion models (which combine sentiment-based and price-based predictions). It challenged the assumption that sentiment-based signals are inferior to price-based predictions in forecasting market movements. Backtesting these signals across three distinct market regimes yielded robust Sharpe ratios of up to 5.07 in trending markets and 3.73 in neutral markets. Our findings demonstrate that language models can serve as effective short-term market predictors. This paradigm shift underscores the untapped capabilities of language models in financial decision-making and opens new avenues for market prediction applications.
comment: 13 pages, 6 figures
♻ ☆ Speech Representation Learning Revisited: The Necessity of Separate Learnable Parameters and Robust Data Augmentation
Speech modeling methods learn one embedding for a fixed segment of speech, typically in between 10-25 ms. The information present in speech can be divided into two categories: "what is being said" (content) and "how it is expressed" (other) and these two are orthogonal in nature causing the optimization algorithm to find a sub-optimal solution if forced to optimize together. This leads to sub-optimal performance in one or all downstream tasks as shown by previous studies. Current self-supervised learning (SSL) methods such as HuBERT are very good at modeling the content information present in speech. Data augmentation improves the performance on tasks which require effective modeling of other information but this leads to a divided capacity of the model. In this work, we conduct a preliminary study to understand the importance of modeling other information using separate learnable parameters. We propose a modified version of HuBERT, termed Other HuBERT (O-HuBERT), to test our hypothesis. Our findings are twofold: first, the O-HuBERT method is able to utilize all layers to build complex features to encode other information; second, a robust data augmentation strategy is essential for learning the information required by tasks that depend on other information and to achieve state-of-the-art (SOTA) performance on the SUPERB benchmark with a similarly sized model (100 million parameters) and pre-training data (960 hours).
♻ ☆ CogCoM: A Visual Language Model with Chain-of-Manipulations Reasoning
Vision-Language Models (VLMs) have demonstrated their broad effectiveness thanks to extensive training in aligning visual instructions to responses. However, such training of conclusive alignment leads models to ignore essential visual reasoning, further resulting in failures in meticulous visual problems and unfaithful responses. Drawing inspiration from human cognition in solving visual problems (e.g., marking, zoom in), this paper introduces Chain of Manipulations, a mechanism that enables VLMs to solve problems step-by-step with evidence. After training, models can solve various visual problems by eliciting intrinsic manipulations (e.g., grounding, zoom in) with results (e.g., boxes, image) actively without involving external tools, while also allowing users to trace error causes. We study the roadmap to implement this mechanism, including (1) a flexible design of manipulations upon extensive analysis, (2) an efficient automated data generation pipeline, (3) a compatible VLM architecture capable of multi-turn multi-image, and (4) a model training process for versatile capabilities. With the design, we also manually annotate 6K high-quality samples for the challenging graphical mathematical problems. Our trained model, \textbf{CogCoM}, equipped with this mechanism with 17B parameters achieves state-of-the-art performance across 9 benchmarks from 4 categories, demonstrating the effectiveness while preserving the interpretability. Our code, model weights, and collected data are publicly available at https://github.com/THUDM/CogCoM.
comment: 21 pages, 10 figures
♻ ☆ Curriculum-style Data Augmentation for LLM-based Metaphor Detection
Recently, utilizing large language models (LLMs) for metaphor detection has achieved promising results. However, these methods heavily rely on the capabilities of closed-source LLMs, which come with relatively high inference costs and latency. To address this, we propose a method for metaphor detection by fine-tuning open-source LLMs, effectively reducing inference costs and latency with a single inference step. Furthermore, metaphor detection suffers from a severe data scarcity problem, which hinders effective fine-tuning of LLMs. To tackle this, we introduce Curriculum-style Data Augmentation (CDA). Specifically, before fine-tuning, we evaluate the training data to identify correctly predicted instances for fine-tuning, while incorrectly predicted instances are used as seed data for data augmentation. This approach enables the model to quickly learn simpler knowledge and progressively acquire more complex knowledge, thereby improving performance incrementally. Experimental results demonstrate that our method achieves state-of-the-art performance across all baselines. Additionally, we provide detailed ablation studies to validate the effectiveness of CDA.
♻ ☆ What is Wrong with Perplexity for Long-context Language Modeling?
Handling long-context inputs is crucial for large language models (LLMs) in tasks such as extended conversations, document summarization, and many-shot in-context learning. While recent approaches have extended the context windows of LLMs and employed perplexity (PPL) as a standard evaluation metric, PPL has proven unreliable for assessing long-context capabilities. The underlying cause of this limitation has remained unclear. In this work, we provide a comprehensive explanation for this issue. We find that PPL overlooks key tokens, which are essential for long-context understanding, by averaging across all tokens and thereby obscuring the true performance of models in long-context scenarios. To address this, we propose \textbf{LongPPL}, a novel metric that focuses on key tokens by employing a long-short context contrastive method to identify them. Our experiments demonstrate that LongPPL strongly correlates with performance on various long-context benchmarks (e.g., Pearson correlation of -0.96), significantly outperforming traditional PPL in predictive accuracy. Additionally, we introduce \textbf{LongCE} (Long-context Cross-Entropy) loss, a re-weighting strategy for fine-tuning that prioritizes key tokens, leading to consistent improvements across diverse benchmarks. In summary, these contributions offer deeper insights into the limitations of PPL and present effective solutions for accurately evaluating and enhancing the long-context capabilities of LLMs. Code is available at https://github.com/PKU-ML/LongPPL.
♻ ☆ Sylber: Syllabic Embedding Representation of Speech from Raw Audio ICLR 2025
Syllables are compositional units of spoken language that efficiently structure human speech perception and production. However, current neural speech representations lack such structure, resulting in dense token sequences that are costly to process. To bridge this gap, we propose a new model, Sylber, that produces speech representations with clean and robust syllabic structure. Specifically, we propose a self-supervised learning (SSL) framework that bootstraps syllabic embeddings by distilling from its own initial unsupervised syllabic segmentation. This results in a highly structured representation of speech features, offering three key benefits: 1) a fast, linear-time syllable segmentation algorithm, 2) efficient syllabic tokenization with an average of 4.27 tokens per second, and 3) novel phonological units suited for efficient spoken language modeling. Our proposed segmentation method is highly robust and generalizes to out-of-domain data and unseen languages without any tuning. By training token-to-speech generative models, fully intelligible speech can be reconstructed from Sylber tokens with a significantly lower bitrate than baseline SSL tokens. This suggests that our model effectively compresses speech into a compact sequence of tokens with minimal information loss. Lastly, we demonstrate that categorical perception-a linguistic phenomenon in speech perception-emerges naturally in Sylber, making the embedding space more categorical and sparse than previous speech features and thus supporting the high efficiency of our tokenization. Together, we present a novel SSL approach for representing speech as syllables, with significant potential for efficient speech tokenization and spoken language modeling.
comment: Accepted at ICLR 2025
♻ ☆ Path-Consistency: Prefix Enhancement for Efficient Inference in LLM
To enhance the reasoning capabilities of large language models (LLMs), self-consistency has gained significant popularity by combining multiple sampling with majority voting. However, the state-of-the-art self-consistency approaches consume substantial computational resources and lead to significant additional time costs due to the multiple sampling. This prevents its full potential from being realized in scenarios where computational resources are critical. To improve the inference efficiency, this paper introduces \textit{path-consistency}, a method that leverages the confidence of answers generated in earlier branches to identify the prefix of the most promising path. By dynamically guiding the generation of subsequent branches based on this prefix, the \textit{path-consistency} mitigates both the errors and redundancies from random or less useful sampling in self-consistency. As a result, it can significantly accelerate the inference process by reducing the number of tokens generated. Our extensive empirical evaluation shows that the \textit{path-consistency} achieves significant acceleration in inference latency ranging from $7.8\%$ to $40.5\%$, while maintaining or even improving task accuracy across different datasets, including mathematical reasoning, common sense reasoning, symbolic reasoning, and code generation.
♻ ☆ Padding Tone: A Mechanistic Analysis of Padding Tokens in T2I Models NAACL 2025
Text-to-image (T2I) diffusion models rely on encoded prompts to guide the image generation process. Typically, these prompts are extended to a fixed length by adding padding tokens before text encoding. Despite being a default practice, the influence of padding tokens on the image generation process has not been investigated. In this work, we conduct the first in-depth analysis of the role padding tokens play in T2I models. We develop two causal techniques to analyze how information is encoded in the representation of tokens across different components of the T2I pipeline. Using these techniques, we investigate when and how padding tokens impact the image generation process. Our findings reveal three distinct scenarios: padding tokens may affect the model's output during text encoding, during the diffusion process, or be effectively ignored. Moreover, we identify key relationships between these scenarios and the model's architecture (cross or self-attention) and its training process (frozen or trained text encoder). These insights contribute to a deeper understanding of the mechanisms of padding tokens, potentially informing future model design and training practices in T2I systems.
comment: Published in: NAACL 2025. Project webpage: https://padding-tone.github.io/
♻ ☆ Empathy Level Alignment via Reinforcement Learning for Empathetic Response Generation
Empathetic response generation, aiming to understand the user's situation and feelings and respond empathically, is crucial in building human-like dialogue systems. Traditional approaches typically employ maximum likelihood estimation as the optimization objective during training, yet fail to align the empathy levels between generated and target responses. To this end, we propose an empathetic response generation framework using reinforcement learning (EmpRL). The framework develops an effective empathy reward function and generates empathetic responses by maximizing the expected reward through reinforcement learning. EmpRL utilizes the pre-trained T5 model as the generator and further fine-tunes it to initialize the policy. To align the empathy levels between generated and target responses within a given context, an empathy reward function containing three empathy communication mechanisms -- emotional reaction, interpretation, and exploration -- is constructed using pre-designed and pre-trained empathy identifiers. During reinforcement learning training, the proximal policy optimization algorithm is used to fine-tune the policy, enabling the generation of empathetic responses. Both automatic and human evaluations demonstrate that the proposed EmpRL framework significantly improves the quality of generated responses, enhances the similarity in empathy levels between generated and target responses, and produces empathetic responses covering both affective and cognitive aspects.
comment: Accepted by IEEE Transactions on Affective Computing
♻ ☆ Efficient Automated Circuit Discovery in Transformers using Contextual Decomposition
Automated mechanistic interpretation research has attracted great interest due to its potential to scale explanations of neural network internals to large models. Existing automated circuit discovery work relies on activation patching or its approximations to identify subgraphs in models for specific tasks (circuits). They often suffer from slow runtime, approximation errors, and specific requirements of metrics, such as non-zero gradients. In this work, we introduce contextual decomposition for transformers (CD-T) to build interpretable circuits in large language models. CD-T can produce circuits of arbitrary level of abstraction, and is the first able to produce circuits as fine-grained as attention heads at specific sequence positions efficiently. CD-T consists of a set of mathematical equations to isolate contribution of model features. Through recursively computing contribution of all nodes in a computational graph of a model using CD-T followed by pruning, we are able to reduce circuit discovery runtime from hours to seconds compared to state-of-the-art baselines. On three standard circuit evaluation datasets (indirect object identification, greater-than comparisons, and docstring completion), we demonstrate that CD-T outperforms ACDC and EAP by better recovering the manual circuits with an average of 97% ROC AUC under low runtimes. In addition, we provide evidence that faithfulness of CD-T circuits is not due to random chance by showing our circuits are 80% more faithful than random circuits of up to 60% of the original model size. Finally, we show CD-T circuits are able to perfectly replicate original models' behavior (faithfulness $ = 1$) using fewer nodes than the baselines for all tasks. Our results underscore the great promise of CD-T for efficient automated mechanistic interpretability, paving the way for new insights into the workings of large language models.
♻ ☆ A Hybrid Transformer Model for Fake News Detection: Leveraging Bayesian Optimization and Bidirectional Recurrent Unit
In this paper, we propose an optimized Transformer model that integrates Bayesian algorithms with a Bidirectional Gated Recurrent Unit (BiGRU), and apply it to fake news classification for the first time. First, we employ the TF-IDF method to extract features from news texts and transform them into numeric representations to facilitate subsequent machine learning tasks. Two sets of experiments are then conducted for fake news detection and classification: one using a Transformer model optimized only with BiGRU, and the other incorporating Bayesian algorithms into the BiGRU-based Transformer. Experimental results show that the BiGRU-optimized Transformer achieves 100% accuracy on the training set and 99.67% on the test set, while the addition of the Bayesian algorithm maintains 100% accuracy on the training set and slightly improves test-set accuracy to 99.73%. This indicates that the Bayesian algorithm boosts model accuracy by 0.06%, further enhancing the detection capability for fake news. Moreover, the proposed algorithm converges rapidly at around the 10th training epoch with accuracy nearing 100%, demonstrating both its effectiveness and its fast classification ability. Overall, the optimized Transformer model, enhanced by the Bayesian algorithm and BiGRU, exhibits excellent continuous learning and detection performance, offering a robust technical means to combat the spread of fake news in the current era of information overload.
♻ ☆ MobA: Multifaceted Memory-Enhanced Adaptive Planning for Efficient Mobile Task Automation NAACL 2025
Existing Multimodal Large Language Model (MLLM)-based agents face significant challenges in handling complex GUI (Graphical User Interface) interactions on devices. These challenges arise from the dynamic and structured nature of GUI environments, which integrate text, images, and spatial relationships, as well as the variability in action spaces across different pages and tasks. To address these limitations, we propose MobA, a novel MLLM-based mobile assistant system. MobA introduces an adaptive planning module that incorporates a reflection mechanism for error recovery and dynamically adjusts plans to align with the real environment contexts and action module's execution capacity. Additionally, a multifaceted memory module provides comprehensive memory support to enhance adaptability and efficiency. We also present MobBench, a dataset designed for complex mobile interactions. Experimental results on MobBench and AndroidArena demonstrate MobA's ability to handle dynamic GUI environments and perform complex mobile task.
comment: NAACL 2025 Demo Track
♻ ☆ Examining Alignment of Large Language Models through Representative Heuristics: The Case of Political Stereotypes ICLR 2025
Examining the alignment of large language models (LLMs) has become increasingly important, e.g., when LLMs fail to operate as intended. This study examines the alignment of LLMs with human values for the domain of politics. Prior research has shown that LLM-generated outputs can include political leanings and mimic the stances of political parties on various issues. However, the extent and conditions under which LLMs deviate from empirical positions are insufficiently examined. To address this gap, we analyze the factors that contribute to LLMs' deviations from empirical positions on political issues, aiming to quantify these deviations and identify the conditions that cause them. Drawing on findings from cognitive science about representativeness heuristics, i.e., situations where humans lean on representative attributes of a target group in a way that leads to exaggerated beliefs, we scrutinize LLM responses through this heuristics' lens. We conduct experiments to determine how LLMs inflate predictions about political parties, which results in stereotyping. We find that while LLMs can mimic certain political parties' positions, they often exaggerate these positions more than human survey respondents do. Also, LLMs tend to overemphasize representativeness more than humans. This study highlights the susceptibility of LLMs to representativeness heuristics, suggesting a potential vulnerability of LLMs that facilitates political stereotyping. We also test prompt-based mitigation strategies, finding that strategies that can mitigate representative heuristics in humans are also effective in reducing the influence of representativeness on LLM-generated responses.
comment: ICLR 2025
♻ ☆ FACT-AUDIT: An Adaptive Multi-Agent Framework for Dynamic Fact-Checking Evaluation of Large Language Models
Large Language Models (LLMs) have significantly advanced the fact-checking studies. However, existing automated fact-checking evaluation methods rely on static datasets and classification metrics, which fail to automatically evaluate the justification production and uncover the nuanced limitations of LLMs in fact-checking. In this work, we introduce FACT-AUDIT, an agent-driven framework that adaptively and dynamically assesses LLMs' fact-checking capabilities. Leveraging importance sampling principles and multi-agent collaboration, FACT-AUDIT generates adaptive and scalable datasets, performs iterative model-centric evaluations, and updates assessments based on model-specific responses. By incorporating justification production alongside verdict prediction, this framework provides a comprehensive and evolving audit of LLMs' factual reasoning capabilities, to investigate their trustworthiness. Extensive experiments demonstrate that FACT-AUDIT effectively differentiates among state-of-the-art LLMs, providing valuable insights into model strengths and limitations in model-centric fact-checking analysis.
♻ ☆ Adaptive In-conversation Team Building for Language Model Agents
Leveraging multiple large language model (LLM) agents has shown to be a promising approach for tackling complex tasks, while the effective design of multiple agents for a particular application remains an art. It is thus intriguing to answer a critical question: Given a task, how can we build a team of LLM agents to solve it effectively? Our new adaptive team-building paradigm offers a flexible solution, realized through a novel agent design named Captain Agent. It dynamically forms and manages teams for each step of a task-solving process, utilizing nested group conversations and reflection to ensure diverse expertise and prevent stereotypical outputs, allowing for a flexible yet structured approach to problem-solving. A comprehensive evaluation across six real-world scenarios demonstrates that Captain Agent significantly outperforms existing multi-agent methods with 21.94% improvement in average accuracy, providing outstanding performance without requiring task-specific prompt engineering. Our exploration of different backbone LLM and cost analysis further shows that Captain Agent can improve the conversation quality of weak LLM and achieve competitive performance with extremely low cost, which illuminates the application of multi-agent systems.
♻ ☆ SeqAR: Jailbreak LLMs with Sequential Auto-Generated Characters NAACL 2025
The widespread applications of large language models (LLMs) have brought about concerns regarding their potential misuse. Although aligned with human preference data before release, LLMs remain vulnerable to various malicious attacks. In this paper, we adopt a red-teaming strategy to enhance LLM safety and introduce SeqAR, a simple yet effective framework to design jailbreak prompts automatically. The SeqAR framework generates and optimizes multiple jailbreak characters and then applies sequential jailbreak characters in a single query to bypass the guardrails of the target LLM. Different from previous work which relies on proprietary LLMs or seed jailbreak templates crafted by human expertise, SeqAR can generate and optimize the jailbreak prompt in a cold-start scenario using open-sourced LLMs without any seed jailbreak templates. Experimental results show that SeqAR achieves attack success rates of 88% and 60% in bypassing the safety alignment of GPT-3.5-1106 and GPT-4, respectively. Furthermore, we extensively evaluate the transferability of the generated templates across different LLMs and held-out malicious requests, while also exploring defense strategies against the jailbreak attack designed by SeqAR.
comment: Accepted by NAACL 2025
♻ ☆ MACPO: Weak-to-Strong Alignment via Multi-Agent Contrastive Preference Optimization ICLR 2025
As large language models (LLMs) are rapidly advancing and achieving near-human capabilities on specific tasks, aligning them with human values is becoming more urgent. In scenarios where LLMs outperform humans, we face a weak-to-strong alignment problem where we need to effectively align strong student LLMs through weak supervision generated by weak teachers. Existing alignment methods mainly focus on strong-to-weak alignment and self-alignment settings, and it is impractical to adapt them to the much harder weak-to-strong alignment setting. To fill this gap, we propose a multi-agent contrastive preference optimization (MACPO) framework. MACPO facilitates weak teachers and strong students to learn from each other by iteratively reinforcing unfamiliar positive behaviors while penalizing familiar negative ones. To get this, we devise a mutual positive behavior augmentation strategy to encourage weak teachers and strong students to learn from each other's positive behavior and further provide higher quality positive behavior for the next iteration. Additionally, we propose a hard negative behavior construction strategy to induce weak teachers and strong students to generate familiar negative behavior by fine-tuning on negative behavioral data. Experimental results on the HH-RLHF and PKU-SafeRLHF datasets, evaluated using both automatic metrics and human judgments, demonstrate that MACPO simultaneously improves the alignment performance of strong students and weak teachers. Moreover, as the number of weak teachers increases, MACPO achieves better weak-to-strong alignment performance through more iteration optimization rounds.
comment: ICLR 2025
♻ ☆ X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale ICLR 2025
Large language models (LLMs) have achieved remarkable success across various NLP tasks with a focus on English due to English-centric pre-training and limited multilingual data. In this work, we focus on the problem of translation, and while some multilingual LLMs claim to support for hundreds of languages, models often fail to provide high-quality responses for mid- and low-resource languages, leading to imbalanced performance heavily skewed in favor of high-resource languages. We introduce **X-ALMA**, a model designed to ensure top-tier performance across 50 diverse languages, regardless of their resource levels. X-ALMA surpasses state-of-the-art open-source multilingual LLMs, such as Aya-101 and Aya-23, in every single translation direction on the FLORES-200 and WMT'23 test datasets according to COMET-22. This is achieved by plug-and-play language-specific module architecture to prevent language conflicts during training and a carefully designed training regimen with novel optimization methods to maximize the translation performance. After the final stage of training regimen, our proposed **A**daptive **R**ejection **P**reference **O**ptimization (**ARPO**) surpasses existing preference optimization methods in translation tasks.
comment: Published as a conference paper at ICLR 2025 (spotlight)
♻ ☆ MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning NAACL 2025
Efficient finetuning of large language models (LLMs) aims to adapt the LLMs with reduced computational and memory cost. Previous LoRA-based approaches initialize the low-rank matrices with Gaussian distribution and zero values while keeping the original weight matrices frozen. However, the trainable model parameters optimized in an unguided subspace might interfere with the well-learned subspace of the pretrained weight matrices. In this paper, we propose MiLoRA, a simple yet effective LLM finetuning approach that only updates the minor singular components of the weight matrix while keeping the principal singular components frozen. It is observed that the minor matrix corresponds to the noisy or long-tail information, while the principal matrix contains important knowledge. The MiLoRA initializes the low-rank matrices within a subspace that is orthogonal to the principal matrix, thus the pretrained knowledge is expected to be well preserved. During finetuning, MiLoRA makes the most use of the less-optimized subspace for learning the labeled dataset. Extensive experiments on commonsense reasoning, math reasoning, instruction following and visual instruction following benchmarks present the superior performance of our method.
comment: This paper has been accepted at NAACL 2025. Code is available at: https://github.com/sufenlp/MiLoRA
♻ ☆ L3Ms -- Lagrange Large Language Models ICLR
Supervised fine-tuning (SFT) and alignment of large language models (LLMs) are key steps in providing a good user experience. However, the concept of an appropriate alignment is inherently application-dependent, and current methods often rely on heuristic choices to drive optimization. In this work, we formulate SFT and alignment as a constrained optimization problem: the LLM is fine-tuned on a task while being required to meet application-specific requirements, without resorting to heuristics. To solve this, we propose Lagrange Large Language Models (L3Ms), which employ logarithmic barriers to enforce the constraints. This approach allows for the customization of L3Ms across diverse applications while avoiding heuristic-driven processes. We experimentally demonstrate the versatility and efficacy of L3Ms in achieving tailored alignments for various applications.
comment: International Conference on Learning Representations (ICLR), 2025
♻ ☆ An Empirical Analysis of Uncertainty in Large Language Model Evaluations ICLR 2025
As LLM-as-a-Judge emerges as a new paradigm for assessing large language models (LLMs), concerns have been raised regarding the alignment, bias, and stability of LLM evaluators. While substantial work has focused on alignment and bias, little research has concentrated on the stability of LLM evaluators. In this paper, we conduct extensive experiments involving 9 widely used LLM evaluators across 2 different evaluation settings to investigate the uncertainty in model-based LLM evaluations. We pinpoint that LLM evaluators exhibit varying uncertainty based on model families and sizes. With careful comparative analyses, we find that employing special prompting strategies, whether during inference or post-training, can alleviate evaluation uncertainty to some extent. By utilizing uncertainty to enhance LLM's reliability and detection capability in Out-Of-Distribution (OOD) data, we further fine-tune an uncertainty-aware LLM evaluator named ConfiLM using a human-annotated fine-tuning set and assess ConfiLM's OOD evaluation ability on a manually designed test set sourced from the 2024 Olympics. Experimental results demonstrate that incorporating uncertainty as additional information during the fine-tuning phase can largely improve the model's evaluation performance in OOD scenarios. The code and data are released at: https://github.com/hasakiXie123/LLM-Evaluator-Uncertainty.
comment: ICLR 2025
♻ ☆ A Survey on Large Language Model based Autonomous Agents
Autonomous agents have long been a prominent research focus in both academic and industry communities. Previous research in this field often focuses on training agents with limited knowledge within isolated environments, which diverges significantly from human learning processes, and thus makes the agents hard to achieve human-like decisions. Recently, through the acquisition of vast amounts of web knowledge, large language models (LLMs) have demonstrated remarkable potential in achieving human-level intelligence. This has sparked an upsurge in studies investigating LLM-based autonomous agents. In this paper, we present a comprehensive survey of these studies, delivering a systematic review of the field of LLM-based autonomous agents from a holistic perspective. More specifically, we first discuss the construction of LLM-based autonomous agents, for which we propose a unified framework that encompasses a majority of the previous work. Then, we present a comprehensive overview of the diverse applications of LLM-based autonomous agents in the fields of social science, natural science, and engineering. Finally, we delve into the evaluation strategies commonly used for LLM-based autonomous agents. Based on the previous studies, we also present several challenges and future directions in this field. To keep track of this field and continuously update our survey, we maintain a repository of relevant references at https://github.com/Paitesanshi/LLM-Agent-Survey.
comment: Correcting several typos, 35 pages, 5 figures, 3 tables
♻ ☆ Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data ICLR 2025
The impressive capabilities of large language models (LLMs) have sparked debate over whether these models genuinely generalize to unseen tasks or predominantly rely on memorizing vast amounts of pretraining data. To explore this issue, we introduce an extended concept of memorization, distributional memorization, which measures the correlation between the LLM output probabilities and the pretraining data frequency. To effectively capture task-specific pretraining data frequency, we propose a novel task-gram language model, which is built by counting the co-occurrence of semantically related $n$-gram pairs from task inputs and outputs in the pretraining corpus. Using the Pythia models trained on the Pile dataset, we evaluate four distinct tasks: machine translation, factual question answering, world knowledge understanding, and math reasoning. Our findings reveal varying levels of memorization, with the strongest effect observed in factual question answering. Furthermore, while model performance improves across all tasks as LLM size increases, only factual question answering shows an increase in memorization, whereas machine translation and reasoning tasks exhibit greater generalization, producing more novel outputs. This study demonstrates that memorization plays a larger role in simpler, knowledge-intensive tasks, while generalization is the key for harder, reasoning-based tasks, providing a scalable method for analyzing large pretraining corpora in greater depth.
comment: Accepted to ICLR 2025
♻ ☆ Mutual Enhancement of Large Language and Reinforcement Learning Models through Bi-Directional Feedback Mechanisms: A Planning Case Study
Large Language Models (LLMs) have demonstrated remarkable capabilities for reinforcement learning (RL) models, such as planning and reasoning capabilities. However, the problems of LLMs and RL model collaboration still need to be solved. In this study, we employ a teacher-student learning framework to tackle these problems, specifically by offering feedback for LLMs using RL models and providing high-level information for RL models with LLMs in a cooperative multi-agent setting. Within this framework, the LLM acts as a teacher, while the RL model acts as a student. The two agents cooperatively assist each other through a process of recursive help, such as "I help you help I help." The LLM agent supplies abstract information to the RL agent, enabling efficient exploration and policy improvement. In turn, the RL agent offers feedback to the LLM agent, providing valuable, real-time information that helps generate more useful tokens. This bi-directional feedback loop promotes optimization, exploration, and mutual improvement for both agents, enabling them to accomplish increasingly challenging tasks. Remarkably, we propose a practical algorithm to address the problem and conduct empirical experiments to evaluate the effectiveness of our method.
♻ ☆ VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents
Retrieval-augmented generation (RAG) is an effective technique that enables large language models (LLMs) to utilize external knowledge sources for generation. However, current RAG systems are solely based on text, rendering it impossible to utilize vision information like layout and images that play crucial roles in real-world multi-modality documents. In this paper, we introduce VisRAG, which tackles this issue by establishing a vision-language model (VLM)-based RAG pipeline. In this pipeline, instead of first parsing the document to obtain text, the document is directly embedded using a VLM as an image and then retrieved to enhance the generation of a VLM. Compared to traditional text-based RAG, VisRAG maximizes the retention and utilization of the data information in the original documents, eliminating the information loss introduced during the parsing process. We collect both open-source and synthetic data to train the retriever in VisRAG and explore a variety of generation methods. Experiments demonstrate that VisRAG outperforms traditional RAG in both the retrieval and generation stages, achieving a 20--40% end-to-end performance gain over traditional text-based RAG pipeline. Further analysis reveals that VisRAG is efficient in utilizing training data and demonstrates strong generalization capability, positioning it as a promising solution for RAG on multi-modality documents. Our code and data are available at https://github.com/openbmb/visrag.
♻ ☆ InstructRAG: Instructing Retrieval-Augmented Generation via Self-Synthesized Rationales ICLR 2025
Retrieval-augmented generation (RAG) has shown promising potential to enhance the accuracy and factuality of language models (LMs). However, imperfect retrievers or noisy corpora can introduce misleading or even erroneous information to the retrieved contents, posing a significant challenge to the generation quality. Existing RAG methods typically address this challenge by directly predicting final answers despite potentially noisy inputs, resulting in an implicit denoising process that is difficult to interpret and verify. On the other hand, the acquisition of explicit denoising supervision is often costly, involving significant human efforts. In this work, we propose InstructRAG, where LMs explicitly learn the denoising process through self-synthesized rationales -- First, we instruct the LM to explain how the ground-truth answer is derived from retrieved documents. Then, these rationales can be used either as demonstrations for in-context learning of explicit denoising or as supervised fine-tuning data to train the model. Compared to standard RAG approaches, InstructRAG requires no additional supervision, allows for easier verification of the predicted answers, and effectively improves generation accuracy. Experiments show InstructRAG consistently outperforms existing RAG methods in both training-free and trainable scenarios, achieving a relative improvement of 8.3% over the best baseline method on average across five knowledge-intensive benchmarks. Extensive analysis indicates that InstructRAG scales well with increased numbers of retrieved documents and consistently exhibits robust denoising ability even in out-of-domain datasets, demonstrating strong generalizability.
comment: ICLR 2025. Code: https://github.com/weizhepei/InstructRAG
♻ ☆ L3Ms - Lagrange Large Language Models ICLR
Supervised fine-tuning (SFT) and alignment of large language models (LLMs) are key steps in providing a good user experience. However, the concept of an appropriate alignment is inherently application-dependent, and current methods often rely on heuristic choices to drive optimization. In this work, we formulate SFT and alignment as a constrained optimization problem: the LLM is fine-tuned on a task while being required to meet application-specific requirements, without resorting to heuristics. To solve this, we propose Lagrange Large Language Models (L3Ms), which employ logarithmic barriers to enforce the constraints. This approach allows for the customization of L3Ms across diverse applications while avoiding heuristic-driven processes. We experimentally demonstrate the versatility and efficacy of L3Ms in achieving tailored alignments for various applications.
comment: International Conference on Learning Representations (ICLR), 2025
Computer Vision and Pattern Recognition 70
♻ ☆ Unsupervised Denoising for Signal-Dependent and Row-Correlated Imaging Noise
Accurate analysis of microscopy images is hindered by the presence of noise. This noise is usually signal-dependent and often additionally correlated along rows or columns of pixels. Current self- and unsupervised denoisers can address signal-dependent noise, but none can reliably remove noise that is also row- or column-correlated. Here, we present the first fully unsupervised deep learning-based denoiser capable of handling imaging noise that is row-correlated as well as signal-dependent. Our approach uses a Variational Autoencoder (VAE) with a specially designed autoregressive decoder. This decoder is capable of modeling row-correlated and signal-dependent noise but is incapable of independently modeling underlying clean signal. The VAE therefore produces latent variables containing only clean signal information, and these are mapped back into image space using a proposed second decoder network. Our method does not require a pre-trained noise model and can be trained from scratch using unpaired noisy data. We benchmark our approach on microscopy datatsets from a range of imaging modalities and sensor types, each with row- or column-correlated, signal-dependent noise, and show that it outperforms existing self- and unsupervised denoisers.
♻ ☆ Eagle: Exploring The Design Space for Multimodal LLMs with Mixture of Encoders
The ability to accurately interpret complex visual information is a crucial topic of multimodal large language models (MLLMs). Recent work indicates that enhanced visual perception significantly reduces hallucinations and improves performance on resolution-sensitive tasks, such as optical character recognition and document analysis. A number of recent MLLMs achieve this goal using a mixture of vision encoders. Despite their success, there is a lack of systematic comparisons and detailed ablation studies addressing critical aspects, such as expert selection and the integration of multiple vision experts. This study provides an extensive exploration of the design space for MLLMs using a mixture of vision encoders and resolutions. Our findings reveal several underlying principles common to various existing strategies, leading to a streamlined yet effective design approach. We discover that simply concatenating visual tokens from a set of complementary vision encoders is as effective as more complex mixing architectures or strategies. We additionally introduce Pre-Alignment to bridge the gap between vision-focused encoders and language tokens, enhancing model coherence. The resulting family of MLLMs, Eagle, surpasses other leading open-source models on major MLLM benchmarks.
comment: Github: https://github.com/NVlabs/Eagle, HuggingFace: https://huggingface.co/NVEagle
♻ ☆ SV-RAG: LoRA-Contextualizing Adaptation of MLLMs for Long Document Understanding ICLR 2025
Multimodal large language models (MLLMs) have recently shown great progress in text-rich image understanding, yet they still struggle with complex, multi-page visually-rich documents. Traditional methods using document parsers for retrieval-augmented generation suffer from performance and efficiency limitations, while directly presenting all pages to MLLMs leads to inefficiencies, especially with lengthy ones. In this work, we present a novel framework named **S**elf-**V**isual **R**etrieval-**A**ugmented **G**eneration (SV-RAG), which can broaden horizons of any MLLM to support long-document understanding. We demonstrate that **MLLMs themselves can be an effective multimodal retriever** to fetch relevant pages and then answer user questions based on these pages. SV-RAG is implemented with two specific MLLM adapters, one for evidence page retrieval and the other for question answering. Empirical results show state-of-the-art performance on public benchmarks, demonstrating the effectiveness of SV-RAG.
comment: Accepted to ICLR 2025
♻ ☆ ALBAR: Adversarial Learning approach to mitigate Biases in Action Recognition ICLR 2025
Bias in machine learning models can lead to unfair decision making, and while it has been well-studied in the image and text domains, it remains underexplored in action recognition. Action recognition models often suffer from background bias (i.e., inferring actions based on background cues) and foreground bias (i.e., relying on subject appearance), which can be detrimental to real-life applications such as autonomous vehicles or assisted living monitoring. While prior approaches have mainly focused on mitigating background bias using specialized augmentations, we thoroughly study both foreground and background bias. We propose ALBAR, a novel adversarial training method that mitigates foreground and background biases without requiring specialized knowledge of the bias attributes. Our framework applies an adversarial cross-entropy loss to the sampled static clip (where all the frames are the same) and aims to make its class probabilities uniform using a proposed entropy maximization loss. Additionally, we introduce a gradient penalty loss for regularization against the debiasing process. We evaluate our method on established background and foreground bias protocols, setting a new state-of-the-art and strongly improving combined debiasing performance by over 12% absolute on HMDB51. Furthermore, we identify an issue of background leakage in the existing UCF101 protocol for bias evaluation which provides a shortcut to predict actions and does not provide an accurate measure of the debiasing capability of a model. We address this issue by proposing more fine-grained segmentation boundaries for the actor, where our method also outperforms existing approaches. Project Page: https://joefioresi718.github.io/ALBAR_webpage/
comment: Accepted to ICLR 2025
♻ ☆ Fréchet Wavelet Distance: A Domain-Agnostic Metric for Image Generation
Modern metrics for generative learning like Fr\'echet Inception Distance (FID) and DINOv2-Fr\'echet Distance (FD-DINOv2) demonstrate impressive performance. However, they suffer from various shortcomings, like a bias towards specific generators and datasets. To address this problem, we propose the Fr\'echet Wavelet Distance (FWD) as a domain-agnostic metric based on the Wavelet Packet Transform ($W_p$). FWD provides a sight across a broad spectrum of frequencies in images with a high resolution, preserving both spatial and textural aspects. Specifically, we use $W_p$ to project generated and real images to the packet coefficient space. We then compute the Fr\'echet distance with the resultant coefficients to evaluate the quality of a generator. This metric is general-purpose and dataset-domain agnostic, as it does not rely on any pre-trained network, while being more interpretable due to its ability to compute Fr\'echet distance per packet, enhancing transparency. We conclude with an extensive evaluation of a wide variety of generators across various datasets that the proposed FWD can generalize and improve robustness to domain shifts and various corruptions compared to other metrics.
♻ ☆ TESGNN: Temporal Equivariant Scene Graph Neural Networks for Efficient and Robust Multi-View 3D Scene Understanding
Scene graphs have proven to be highly effective for various scene understanding tasks due to their compact and explicit representation of relational information. However, current methods often overlook the critical importance of preserving symmetry when generating scene graphs from 3D point clouds, which can lead to reduced accuracy and robustness, particularly when dealing with noisy, multi-view data. Furthermore, a major limitation of prior approaches is the lack of temporal modeling to capture time-dependent relationships among dynamically evolving entities in a scene. To address these challenges, we propose Temporal Equivariant Scene Graph Neural Network (TESGNN), consisting of two key components: (1) an Equivariant Scene Graph Neural Network (ESGNN), which extracts information from 3D point clouds to generate scene graph while preserving crucial symmetry properties, and (2) a Temporal Graph Matching Network, which fuses scene graphs generated by ESGNN across multiple time sequences into a unified global representation using an approximate graph-matching algorithm. Our combined architecture TESGNN outperforms current state-of-the-art methods in scene graph generation, achieving higher accuracy and faster training convergence. Moreover, we show that leveraging the symmetry-preserving property produces a more stable and accurate global scene representation compared to existing approaches. Last but not least, it is computationally efficient and easily implementable using existing frameworks, making it well-suited for real-time applications in robotics and computer vision. This approach paves the way for more robust and scalable solutions to complex multi-view scene understanding challenges. Our source code is publicly available at: https://github.com/HySonLab/TESGraph
comment: arXiv admin note: text overlap with arXiv:2407.00609
♻ ☆ Learning General-Purpose Biomedical Volume Representations using Randomized Synthesis ICLR 2025
Current volumetric biomedical foundation models struggle to generalize as public 3D datasets are small and do not cover the broad diversity of medical procedures, conditions, anatomical regions, and imaging protocols. We address this by creating a representation learning method that instead anticipates strong domain shifts at training time itself. We first propose a data engine that synthesizes highly variable training samples that would enable generalization to new biomedical contexts. To then train a single 3D network for any voxel-level task, we develop a contrastive learning method that pretrains the network to be stable against nuisance imaging variation simulated by the data engine, a key inductive bias for generalization. This network's features can be used as robust representations of input images for downstream tasks and its weights provide a strong, dataset-agnostic initialization for finetuning on new datasets. As a result, we set new standards across both multimodality registration and few-shot segmentation, a first for any 3D biomedical vision model, all without (pre-)training on any existing dataset of real images.
comment: ICLR 2025: International Conference on Learning Representations. Code and model weights available at https://github.com/neel-dey/anatomix. Keywords: synthetic data, representation learning, medical image analysis, image registration, image segmentation
♻ ☆ Tri-Clustering: A Multi-views Tri-level Information Fusion Context Clustering Framework for Localization and Classification in Mammography
Breast cancer is a significant global health issue, and the diagnosis of breast imaging has always been challenging. Mammography images typically have extremely high resolution, with lesions occupying only a very small area. Down-sampling in neural networks can easily lead to the loss of microcalcifications or subtle structures, making it difficult for traditional neural network architectures to address these issues. To tackle these challenges, we propose a Context Clustering Network with triple information fusion. Firstly, compared to CNNs or transformers, we find that Context clustering methods (1) are more computationally efficient and (2) can more easily associate structural or pathological features, making them suitable for the clinical tasks of mammography. Secondly, we propose a triple information fusion mechanism that integrates global information, feature-based local information, and patch-based local information. The proposed approach is rigorously evaluated on two public datasets, Vindr-Mammo and CBIS-DDSM, using five independent splits to ensure statistical robustness. Our method achieves an AUC of 0.828 on Vindr-Mammo and 0.805 on CBIS-DDSM, outperforming the next best method by 3.1% and 2.4%, respectively. These improvements are statistically significant (p<0.05), underscoring the benefits of Context Clustering Network with triple information fusion. Overall, our Context Clustering framework demonstrates strong potential as a scalable and cost-effective solution for large-scale mammography screening, enabling more efficient and accurate breast cancer detection. Access to our method is available at https://github.com/Sohyu1/Mammo_Clustering.
comment: 10 pages, 6 figures
♻ ☆ FedBiP: Heterogeneous One-Shot Federated Learning with Personalized Latent Diffusion Models CVPR 2025
One-Shot Federated Learning (OSFL), a special decentralized machine learning paradigm, has recently gained significant attention. OSFL requires only a single round of client data or model upload, which reduces communication costs and mitigates privacy threats compared to traditional FL. Despite these promising prospects, existing methods face challenges due to client data heterogeneity and limited data quantity when applied to real-world OSFL systems. Recently, Latent Diffusion Models (LDM) have shown remarkable advancements in synthesizing high-quality images through pretraining on large-scale datasets, thereby presenting a potential solution to overcome these issues. However, directly applying pretrained LDM to heterogeneous OSFL results in significant distribution shifts in synthetic data, leading to performance degradation in classification models trained on such data. This issue is particularly pronounced in rare domains, such as medical imaging, which are underrepresented in LDM's pretraining data. To address this challenge, we propose Federated Bi-Level Personalization (FedBiP), which personalizes the pretrained LDM at both instance-level and concept-level. Hereby, FedBiP synthesizes images following the client's local data distribution without compromising the privacy regulations. FedBiP is also the first approach to simultaneously address feature space heterogeneity and client data scarcity in OSFL. Our method is validated through extensive experiments on three OSFL benchmarks with feature space heterogeneity, as well as on challenging medical and satellite image datasets with label heterogeneity. The results demonstrate the effectiveness of FedBiP, which substantially outperforms other OSFL methods.
comment: CVPR 2025
♻ ☆ Synthesizing Physically Plausible Human Motions in 3D Scenes 3DV 2024
We present a physics-based character control framework for synthesizing human-scene interactions. Recent advances adopt physics simulation to mitigate artifacts produced by data-driven kinematic approaches. However, existing physics-based methods mainly focus on single-object environments, resulting in limited applicability in realistic 3D scenes with multi-objects. To address such challenges, we propose a framework that enables physically simulated characters to perform long-term interaction tasks in diverse, cluttered, and unseen 3D scenes. The key idea is to decouple human-scene interactions into two fundamental processes, Interacting and Navigating, which motivates us to construct two reusable Controllers, namely InterCon and NavCon. Specifically, InterCon uses two complementary policies to enable characters to enter or leave the interacting state with a particular object (e.g., sitting on a chair or getting up). To realize navigation in cluttered environments, we introduce NavCon, where a trajectory following policy enables characters to track pre-planned collision-free paths. Benefiting from the divide and conquer strategy, we can train all policies in simple environments and directly apply them in complex multi-object scenes through coordination from a rule-based scheduler. Video and code are available at https://github.com/liangpan99/InterScene.
comment: 3DV 2024 version
♻ ☆ Unleashing the Potential of Vision-Language Pre-Training for 3D Zero-Shot Lesion Segmentation via Mask-Attribute Alignment ICLR 2025
Recent advancements in medical vision-language pre-training models have driven significant progress in zero-shot disease recognition. However, transferring image-level knowledge to pixel-level tasks, such as lesion segmentation in 3D CT scans, remains a critical challenge. Due to the complexity and variability of pathological visual characteristics, existing methods struggle to align fine-grained lesion features not encountered during training with disease-related textual representations. In this paper, we present Malenia, a novel multi-scale lesion-level mask-attribute alignment framework, specifically designed for 3D zero-shot lesion segmentation. Malenia improves the compatibility between mask representations and their associated elemental attributes, explicitly linking the visual features of unseen lesions with the extensible knowledge learned from previously seen ones. Furthermore, we design a Cross-Modal Knowledge Injection module to enhance both visual and textual features with mutually beneficial information, effectively guiding the generation of segmentation results. Comprehensive experiments across three datasets and 12 lesion categories validate the superior performance of Malenia.
comment: Accepted as ICLR 2025 conference paper
♻ ☆ Bidirectional Consistency Models ICML 2024
Diffusion models (DMs) are capable of generating remarkably high-quality samples by iteratively denoising a random vector, a process that corresponds to moving along the probability flow ordinary differential equation (PF ODE). Interestingly, DMs can also invert an input image to noise by moving backward along the PF ODE, a key operation for downstream tasks such as interpolation and image editing. However, the iterative nature of this process restricts its speed, hindering its broader application. Recently, Consistency Models (CMs) have emerged to address this challenge by approximating the integral of the PF ODE, largely reducing the number of iterations. Yet, the absence of an explicit ODE solver complicates the inversion process. To resolve this, we introduce Bidirectional Consistency Model (BCM), which learns a single neural network that enables both forward and backward traversal along the PF ODE, efficiently unifying generation and inversion tasks within one framework. We can train BCM from scratch or tune it using a pretrained consistency model, which reduces the training cost and increases scalability. We demonstrate that BCM enables one-step generation and inversion while also allowing the use of additional steps to enhance generation quality or reduce reconstruction error. We further showcase BCM's capability in downstream tasks, such as interpolation and inpainting. Our code and weights are available at https://github.com/Mosasaur5526/BCM-iCT-torch.
comment: 39 pages, 27 figures; a shorter version of this paper was acceppted at the ICML 2024 Workshop on Structured Probabilistic Inference & Generative Modeling
♻ ☆ Video-Foley: Two-Stage Video-To-Sound Generation via Temporal Event Condition For Foley Sound
Foley sound synthesis is crucial for multimedia production, enhancing user experience by synchronizing audio and video both temporally and semantically. Recent studies on automating this labor-intensive process through video-to-sound generation face significant challenges. Systems lacking explicit temporal features suffer from poor alignment and controllability, while timestamp-based models require costly and subjective human annotation. We propose Video-Foley, a video-to-sound system using Root Mean Square (RMS) as an intuitive condition with semantic timbre prompts (audio or text). RMS, a frame-level intensity envelope closely related to audio semantics, acts as a temporal event feature to guide audio generation from video. The annotation-free self-supervised learning framework consists of two stages, Video2RMS and RMS2Sound, incorporating novel ideas including RMS discretization and RMS-ControlNet with a pretrained text-to-audio model. Our extensive evaluation shows that Video-Foley achieves state-of-the-art performance in audio-visual alignment and controllability for sound timing, intensity, timbre, and nuance. Source code, model weights and demos are available on our companion website. (https://jnwnlee.github.io/video-foley-demo)
♻ ☆ LLaVA-Mini: Efficient Image and Video Large Multimodal Models with One Vision Token ICLR 2025
The advent of real-time large multimodal models (LMMs) like GPT-4o has sparked considerable interest in efficient LMMs. LMM frameworks typically encode visual inputs into vision tokens (continuous representations) and integrate them and textual instructions into the context of large language models (LLMs), where large-scale parameters and numerous context tokens (predominantly vision tokens) result in substantial computational overhead. Previous efforts towards efficient LMMs always focus on replacing the LLM backbone with smaller models, while neglecting the crucial issue of token quantity. In this paper, we introduce LLaVA-Mini, an efficient LMM with minimal vision tokens. To achieve a high compression ratio of vision tokens while preserving visual information, we first analyze how LMMs understand vision tokens and find that most vision tokens only play a crucial role in the early layers of LLM backbone, where they mainly fuse visual information into text tokens. Building on this finding, LLaVA-Mini introduces modality pre-fusion to fuse visual information into text tokens in advance, thereby facilitating the extreme compression of vision tokens fed to LLM backbone into one token. LLaVA-Mini is a unified large multimodal model that can support the understanding of images, high-resolution images, and videos in an efficient manner. Experiments across 11 image-based and 7 video-based benchmarks demonstrate that LLaVA-Mini outperforms LLaVA-v1.5 with just 1 vision token instead of 576. Efficiency analyses reveal that LLaVA-Mini can reduce FLOPs by 77%, deliver low-latency responses within 40 milliseconds, and process over 10,000 frames of video on the GPU hardware with 24GB of memory.
comment: Accepted to ICLR 2025. Code: https://github.com/ictnlp/LLaVA-Mini Model: https://huggingface.co/ICTNLP/llava-mini-llama-3.1-8b
♻ ☆ Drag Your Gaussian: Effective Drag-Based Editing with Score Distillation for 3D Gaussian Splatting
Recent advancements in 3D scene editing have been propelled by the rapid development of generative models. Existing methods typically utilize generative models to perform text-guided editing on 3D representations, such as 3D Gaussian Splatting (3DGS). However, these methods are often limited to texture modifications and fail when addressing geometric changes, such as editing a character's head to turn around. Moreover, such methods lack accurate control over the spatial position of editing results, as language struggles to precisely describe the extent of edits. To overcome these limitations, we introduce DYG, an effective 3D drag-based editing method for 3D Gaussian Splatting. It enables users to conveniently specify the desired editing region and the desired dragging direction through the input of 3D masks and pairs of control points, thereby enabling precise control over the extent of editing. DYG integrates the strengths of the implicit triplane representation to establish the geometric scaffold of the editing results, effectively overcoming suboptimal editing outcomes caused by the sparsity of 3DGS in the desired editing regions. Additionally, we incorporate a drag-based Latent Diffusion Model into our method through the proposed Drag-SDS loss function, enabling flexible, multi-view consistent, and fine-grained editing. Extensive experiments demonstrate that DYG conducts effective drag-based editing guided by control point prompts, surpassing other baselines in terms of editing effect and quality, both qualitatively and quantitatively. Visit our project page at https://quyans.github.io/Drag-Your-Gaussian.
comment: Visit our project page at https://quyans.github.io/Drag-Your-Gaussian
♻ ☆ Audio-Visual Instance Segmentation CVPR 2025
In this paper, we propose a new multi-modal task, termed audio-visual instance segmentation (AVIS), which aims to simultaneously identify, segment and track individual sounding object instances in audible videos. To facilitate this research, we introduce a high-quality benchmark named AVISeg, containing over 90K instance masks from 26 semantic categories in 926 long videos. Additionally, we propose a strong baseline model for this task. Our model first localizes sound source within each frame, and condenses object-specific contexts into concise tokens. Then it builds long-range audio-visual dependencies between these tokens using window-based attention, and tracks sounding objects among the entire video sequences. Extensive experiments reveal that our method performs best on AVISeg, surpassing the existing methods from related tasks. We further conduct the evaluation on several multi-modal large models. Unfortunately, they exhibits subpar performance on instance-level sound source localization and temporal perception. We expect that AVIS will inspire the community towards a more comprehensive multi-modal understanding. Dataset and code is available at https://github.com/ruohaoguo/avis.
comment: Accepted by CVPR 2025
♻ ☆ Evaluating Low-Resource Lane Following Algorithms for Compute-Constrained Automated Vehicles
Reliable lane-following is essential for automated and assisted driving, yet existing solutions often rely on models that require extensive computational resources, limiting their deployment in compute-constrained vehicles. We evaluate five low-resource lane-following algorithms designed for real-time operation on vehicles with limited computing resources. Performance was assessed through simulation and deployment on real drive-by-wire electric vehicles, with evaluation metrics including reliability, comfort, speed, and adaptability. The top-performing methods used unsupervised learning to detect and separate lane lines with processing time under 10 ms per frame, outperforming compute-intensive and poor generalizing deep learning approaches. These approaches demonstrated robustness across lighting conditions, road textures, and lane geometries. The findings highlight the potential for efficient lane detection approaches to enhance the accessibility and reliability of autonomous vehicle technologies. Reducing computing requirements enables lane keeping to be widely deployed in vehicles as part of lower-level automation, including active safety systems.
comment: Supported by the National Science Foundation under Grants No. 2150292 and 2150096
♻ ☆ SPARTUN3D: Situated Spatial Understanding of 3D World in Large Language Models
Integrating the 3D world into large language models (3D-based LLMs) has been a promising research direction for 3D scene understanding. However, current 3D-based LLMs fall short in situated understanding due to two key limitations: 1) existing 3D datasets are constructed from a global perspective of the 3D scenes and lack situated context. 2) the architectures of existing 3D-based LLMs lack explicit alignment between the spatial representations of 3D scenes and natural language, limiting their performance in tasks requiring precise spatial reasoning. We address these issues by introducing a scalable situated 3D dataset, named Spartun3D, that incorporates various situated spatial reasoning tasks. Furthermore, we propose Spartun3D-LLM, built on an existing 3D-based LLM but integrated with a novel situated spatial alignment module, aiming to enhance the alignment between 3D visual representations and their corresponding textual descriptions. Experimental results demonstrate that both our proposed dataset and alignment module significantly enhance the situated spatial understanding of 3D-based LLMs.
♻ ☆ HMD^2: Environment-aware Motion Generation from Single Egocentric Head-Mounted Device 3DV 2025
This paper investigates the generation of realistic full-body human motion using a single head-mounted device with an outward-facing color camera and the ability to perform visual SLAM. To address the ambiguity of this setup, we present HMD^2, a novel system that balances motion reconstruction and generation. From a reconstruction standpoint, it aims to maximally utilize the camera streams to produce both analytical and learned features, including head motion, SLAM point cloud, and image embeddings. On the generative front, HMD^2 employs a multi-modal conditional motion diffusion model with a Transformer backbone to maintain temporal coherence of generated motions, and utilizes autoregressive inpainting to facilitate online motion inference with minimal latency (0.17 seconds). We show that our system provides an effective and robust solution that scales to a diverse dataset of over 200 hours of motion in complex indoor and outdoor environments.
comment: International Conference on 3D Vision 2025 (3DV 2025)
♻ ☆ Rethinking Audio-Visual Adversarial Vulnerability from Temporal and Modality Perspectives ICLR 2025
While audio-visual learning equips models with a richer understanding of the real world by leveraging multiple sensory modalities, this integration also introduces new vulnerabilities to adversarial attacks. In this paper, we present a comprehensive study of the adversarial robustness of audio-visual models, considering both temporal and modality-specific vulnerabilities. We propose two powerful adversarial attacks: 1) a temporal invariance attack that exploits the inherent temporal redundancy across consecutive time segments and 2) a modality misalignment attack that introduces incongruence between the audio and visual modalities. These attacks are designed to thoroughly assess the robustness of audio-visual models against diverse threats. Furthermore, to defend against such attacks, we introduce a novel audio-visual adversarial training framework. This framework addresses key challenges in vanilla adversarial training by incorporating efficient adversarial perturbation crafting tailored to multi-modal data and an adversarial curriculum strategy. Extensive experiments in the Kinetics-Sounds dataset demonstrate that our proposed temporal and modality-based attacks in degrading model performance can achieve state-of-the-art performance, while our adversarial training defense largely improves the adversarial robustness as well as the adversarial training efficiency.
comment: Accepted by ICLR 2025
♻ ☆ Tracking objects that change in appearance with phase synchrony
Objects we encounter often change appearance as we interact with them. Changes in illumination (shadows), object pose, or the movement of non-rigid objects can drastically alter available image features. How do biological visual systems track objects as they change? One plausible mechanism involves attentional mechanisms for reasoning about the locations of objects independently of their appearances -- a capability that prominent neuroscience theories have associated with computing through neural synchrony. Here, we describe a novel deep learning circuit that can learn to precisely control attention to features separately from their location in the world through neural synchrony: the complex-valued recurrent neural network (CV-RNN). Next, we compare object tracking in humans, the CV-RNN, and other deep neural networks (DNNs), using FeatureTracker: a large-scale challenge that asks observers to track objects as their locations and appearances change in precisely controlled ways. While humans effortlessly solved FeatureTracker, state-of-the-art DNNs did not. In contrast, our CV-RNN behaved similarly to humans on the challenge, providing a computational proof-of-concept for the role of phase synchronization as a neural substrate for tracking appearance-morphing objects as they move about.
♻ ☆ A Dual-Purpose Framework for Backdoor Defense and Backdoor Amplification in Diffusion Models
Diffusion models have emerged as state-of-the-art generative frameworks, excelling in producing high-quality multi-modal samples. However, recent studies have revealed their vulnerability to backdoor attacks, where backdoored models generate specific, undesirable outputs called backdoor target (e.g., harmful images) when a pre-defined trigger is embedded to their inputs. In this paper, we propose PureDiffusion, a dual-purpose framework that simultaneously serves two contrasting roles: backdoor defense and backdoor attack amplification. For defense, we introduce two novel loss functions to invert backdoor triggers embedded in diffusion models. The first leverages trigger-induced distribution shifts across multiple timesteps of the diffusion process, while the second exploits the denoising consistency effect when a backdoor is activated. Once an accurate trigger inversion is achieved, we develop a backdoor detection method that analyzes both the inverted trigger and the generated backdoor targets to identify backdoor attacks. In terms of attack amplification with the role of an attacker, we describe how our trigger inversion algorithm can be used to reinforce the original trigger embedded in the backdoored diffusion model. This significantly boosts attack performance while reducing the required backdoor training time. Experimental results demonstrate that PureDiffusion achieves near-perfect detection accuracy, outperforming existing defenses by a large margin, particularly against complex trigger patterns. Additionally, in an attack scenario, our attack amplification approach elevates the attack success rate (ASR) of existing backdoor attacks to nearly 100\% while reducing training time by up to 20x.
♻ ☆ Test-Time Adaptation for Combating Missing Modalities in Egocentric Videos
Understanding videos that contain multiple modalities is crucial, especially in egocentric videos, where combining various sensory inputs significantly improves tasks like action recognition and moment localization. However, real-world applications often face challenges with incomplete modalities due to privacy concerns, efficiency needs, or hardware issues. Current methods, while effective, often necessitate retraining the model entirely to handle missing modalities, making them computationally intensive, particularly with large training datasets. In this study, we propose a novel approach to address this issue at test time without requiring retraining. We frame the problem as a test-time adaptation task, where the model adjusts to the available unlabeled data at test time. Our method, MiDl~(Mutual information with self-Distillation), encourages the model to be insensitive to the specific modality source present during testing by minimizing the mutual information between the prediction and the available modality. Additionally, we incorporate self-distillation to maintain the model's original performance when both modalities are available. MiDl represents the first self-supervised, online solution for handling missing modalities exclusively at test time. Through experiments with various pretrained models and datasets, MiDl demonstrates substantial performance improvement without the need for retraining.
♻ ☆ DIPSER: A Dataset for In-Person Student Engagement Recognition in the Wild
In this paper, a novel dataset is introduced, designed to assess student attention within in-person classroom settings. This dataset encompasses RGB camera data, featuring multiple cameras per student to capture both posture and facial expressions, in addition to smartwatch sensor data for each individual. This dataset allows machine learning algorithms to be trained to predict attention and correlate it with emotion. A comprehensive suite of attention and emotion labels for each student is provided, generated through self-reporting as well as evaluations by four different experts. Our dataset uniquely combines facial and environmental camera data, smartwatch metrics, and includes underrepresented ethnicities in similar datasets, all within in-the-wild, in-person settings, making it the most comprehensive dataset of its kind currently available. The dataset presented offers an extensive and diverse collection of data pertaining to student interactions across different educational contexts, augmented with additional metadata from other tools. This initiative addresses existing deficiencies by offering a valuable resource for the analysis of student attention and emotion in face-to-face lessons.
♻ ☆ MOVE: Effective and Harmless Ownership Verification via Embedded External Features AAAI 2022
Currently, deep neural networks (DNNs) are widely adopted in different applications. Despite its commercial values, training a well-performing DNN is resource-consuming. Accordingly, the well-trained model is valuable intellectual property for its owner. However, recent studies revealed the threats of model stealing, where the adversaries can obtain a function-similar copy of the victim model, even when they can only query the model. In this paper, we propose an effective and harmless model ownership verification (MOVE) to defend against different types of model stealing simultaneously, without introducing new security risks. In general, we conduct the ownership verification by verifying whether a suspicious model contains the knowledge of defender-specified external features. Specifically, we embed the external features by modifying a few training samples with style transfer. We then train a meta-classifier to determine whether a model is stolen from the victim. This approach is inspired by the understanding that the stolen models should contain the knowledge of features learned by the victim model. In particular, \revision{we develop our MOVE method under both white-box and black-box settings and analyze its theoretical foundation to provide comprehensive model protection.} Extensive experiments on benchmark datasets verify the effectiveness of our method and its resistance to potential adaptive attacks. The codes for reproducing the main experiments of our method are available at https://github.com/THUYimingLi/MOVE.
comment: This paper has been accepted by IEEE TPAMI 2025. It is the journal extension of our conference paper in AAAI 2022 (https://ojs.aaai.org/index.php/AAAI/article/view/20036). 18 pages
♻ ☆ K-LoRA: Unlocking Training-Free Fusion of Any Subject and Style LoRAs CVPR 2025
Recent studies have explored combining different LoRAs to jointly generate learned style and content. However, existing methods either fail to effectively preserve both the original subject and style simultaneously or require additional training. In this paper, we argue that the intrinsic properties of LoRA can effectively guide diffusion models in merging learned subject and style. Building on this insight, we propose K-LoRA, a simple yet effective training-free LoRA fusion approach. In each attention layer, K-LoRA compares the Top-K elements in each LoRA to be fused, determining which LoRA to select for optimal fusion. This selection mechanism ensures that the most representative features of both subject and style are retained during the fusion process, effectively balancing their contributions. Experimental results demonstrate that the proposed method effectively integrates the subject and style information learned by the original LoRAs, outperforming state-of-the-art training-based approaches in both qualitative and quantitative results.
comment: CVPR 2025, Project page: https://k-lora.github.io/K-LoRA.io/
♻ ☆ Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations ICLR 2025
To gain insight into the mechanisms behind machine learning methods, it is crucial to establish connections among the features describing data points. However, these correlations often exhibit a high-dimensional and strongly nonlinear nature, which makes them challenging to detect using standard methods. This paper exploits the entanglement between intrinsic dimensionality and correlation to propose a metric that quantifies the (potentially nonlinear) correlation between high-dimensional manifolds. We first validate our method on synthetic data in controlled environments, showcasing its advantages and drawbacks compared to existing techniques. Subsequently, we extend our analysis to large-scale applications in neural network representations. Specifically, we focus on latent representations of multimodal data, uncovering clear correlations between paired visual and textual embeddings, whereas existing methods struggle significantly in detecting similarity. Our results indicate the presence of highly nonlinear correlation patterns between latent manifolds.
comment: Accepted at ICLR 2025
♻ ☆ Q-Bench-Video: Benchmarking the Video Quality Understanding of LMMs
With the rising interest in research on Large Multi-modal Models (LMMs) for video understanding, many studies have emphasized general video comprehension capabilities, neglecting the systematic exploration into video quality understanding. To address this oversight, we introduce Q-Bench-Video in this paper, a new benchmark specifically designed to evaluate LMMs' proficiency in discerning video quality. a) To ensure video source diversity, Q-Bench-Video encompasses videos from natural scenes, AI-generated Content (AIGC), and Computer Graphics (CG). b) Building on the traditional multiple-choice questions format with the Yes-or-No and What-How categories, we include Open-ended questions to better evaluate complex scenarios. Additionally, we incorporate the video pair quality comparison question to enhance comprehensiveness. c) Beyond the traditional Technical, Aesthetic, and Temporal distortions, we have expanded our evaluation aspects to include the dimension of AIGC distortions, which addresses the increasing demand for video generation. Finally, we collect a total of 2,378 question-answer pairs and test them on 12 open-source & 5 proprietary LMMs. Our findings indicate that while LMMs have a foundational understanding of video quality, their performance remains incomplete and imprecise, with a notable discrepancy compared to human performance. Through Q-Bench-Video, we seek to catalyze community interest, stimulate further research, and unlock the untapped potential of LMMs to close the gap in video quality understanding.
♻ ☆ HyperFace: Generating Synthetic Face Recognition Datasets by Exploring Face Embedding Hypersphere ICLR 2025
Face recognition datasets are often collected by crawling Internet and without individuals' consents, raising ethical and privacy concerns. Generating synthetic datasets for training face recognition models has emerged as a promising alternative. However, the generation of synthetic datasets remains challenging as it entails adequate inter-class and intra-class variations. While advances in generative models have made it easier to increase intra-class variations in face datasets (such as pose, illumination, etc.), generating sufficient inter-class variation is still a difficult task. In this paper, we formulate the dataset generation as a packing problem on the embedding space (represented on a hypersphere) of a face recognition model and propose a new synthetic dataset generation approach, called HyperFace. We formalize our packing problem as an optimization problem and solve it with a gradient descent-based approach. Then, we use a conditional face generator model to synthesize face images from the optimized embeddings. We use our generated datasets to train face recognition models and evaluate the trained models on several benchmarking real datasets. Our experimental results show that models trained with HyperFace achieve state-of-the-art performance in training face recognition using synthetic datasets.
comment: Accepted in ICLR 2025
♻ ☆ Improved Baselines with Synchronized Encoding for Universal Medical Image Segmentation
Large foundation models, known for their strong zero-shot generalization capabilities, can be applied to a wide range of downstream tasks. However, developing foundation models for medical image segmentation poses a significant challenge due to the domain gap between natural and medical images. While fine-tuning techniques based on the Segment Anything Model (SAM) have been explored, they primarily focus on scaling up data or refining inference strategies without incorporating domain-specific architectural designs, limiting their zero-shot performance. To optimize segmentation performance under standard inference settings and provide a strong baseline for future research, we introduce SyncSAM, which employs a synchronized dual-branch encoder that integrates convolution and Transformer features in a synchronized manner to enhance medical image encoding, and a multi-scale dual-branch decoder to preserve image details. SyncSAM is trained on two of the largest medical image segmentation datasets, SA-Med2D-20M and IMed-361M, resulting in a series of pre-trained models for universal medical image segmentation. Experimental results demonstrate that SyncSAM not only achieves state-of-the-art performance on test sets but also exhibits strong zero-shot capabilities on unseen datasets. The code and model weights are available at https://github.com/Hhankyangg/SyncSAM.
♻ ☆ StochSync: Stochastic Diffusion Synchronization for Image Generation in Arbitrary Spaces ICLR 2025
We propose a zero-shot method for generating images in arbitrary spaces (e.g., a sphere for 360{\deg} panoramas and a mesh surface for texture) using a pretrained image diffusion model. The zero-shot generation of various visual content using a pretrained image diffusion model has been explored mainly in two directions. First, Diffusion Synchronization-performing reverse diffusion processes jointly across different projected spaces while synchronizing them in the target space-generates high-quality outputs when enough conditioning is provided, but it struggles in its absence. Second, Score Distillation Sampling-gradually updating the target space data through gradient descent-results in better coherence but often lacks detail. In this paper, we reveal for the first time the interconnection between these two methods while highlighting their differences. To this end, we propose StochSync, a novel approach that combines the strengths of both, enabling effective performance with weak conditioning. Our experiments demonstrate that StochSync provides the best performance in 360{\deg} panorama generation (where image conditioning is not given), outperforming previous finetuning-based methods, and also delivers comparable results in 3D mesh texturing (where depth conditioning is provided) with previous methods.
comment: Project page: https://stochsync.github.io/ (ICLR 2025)
♻ ☆ DartControl: A Diffusion-Based Autoregressive Motion Model for Real-Time Text-Driven Motion Control ICLR
Text-conditioned human motion generation, which allows for user interaction through natural language, has become increasingly popular. Existing methods typically generate short, isolated motions based on a single input sentence. However, human motions are continuous and can extend over long periods, carrying rich semantics. Creating long, complex motions that precisely respond to streams of text descriptions, particularly in an online and real-time setting, remains a significant challenge. Furthermore, incorporating spatial constraints into text-conditioned motion generation presents additional challenges, as it requires aligning the motion semantics specified by text descriptions with geometric information, such as goal locations and 3D scene geometry. To address these limitations, we propose DartControl, in short DART, a Diffusion-based Autoregressive motion primitive model for Real-time Text-driven motion control. Our model effectively learns a compact motion primitive space jointly conditioned on motion history and text inputs using latent diffusion models. By autoregressively generating motion primitives based on the preceding history and current text input, DART enables real-time, sequential motion generation driven by natural language descriptions. Additionally, the learned motion primitive space allows for precise spatial motion control, which we formulate either as a latent noise optimization problem or as a Markov decision process addressed through reinforcement learning. We present effective algorithms for both approaches, demonstrating our model's versatility and superior performance in various motion synthesis tasks. Experiments show our method outperforms existing baselines in motion realism, efficiency, and controllability. Video results are available on the project page: https://zkf1997.github.io/DART/.
comment: Updated ICLR camera ready version
♻ ☆ CogCoM: A Visual Language Model with Chain-of-Manipulations Reasoning
Vision-Language Models (VLMs) have demonstrated their broad effectiveness thanks to extensive training in aligning visual instructions to responses. However, such training of conclusive alignment leads models to ignore essential visual reasoning, further resulting in failures in meticulous visual problems and unfaithful responses. Drawing inspiration from human cognition in solving visual problems (e.g., marking, zoom in), this paper introduces Chain of Manipulations, a mechanism that enables VLMs to solve problems step-by-step with evidence. After training, models can solve various visual problems by eliciting intrinsic manipulations (e.g., grounding, zoom in) with results (e.g., boxes, image) actively without involving external tools, while also allowing users to trace error causes. We study the roadmap to implement this mechanism, including (1) a flexible design of manipulations upon extensive analysis, (2) an efficient automated data generation pipeline, (3) a compatible VLM architecture capable of multi-turn multi-image, and (4) a model training process for versatile capabilities. With the design, we also manually annotate 6K high-quality samples for the challenging graphical mathematical problems. Our trained model, \textbf{CogCoM}, equipped with this mechanism with 17B parameters achieves state-of-the-art performance across 9 benchmarks from 4 categories, demonstrating the effectiveness while preserving the interpretability. Our code, model weights, and collected data are publicly available at https://github.com/THUDM/CogCoM.
comment: 21 pages, 10 figures
♻ ☆ CLIPure: Purification in Latent Space via CLIP for Adversarially Robust Zero-Shot Classification ICLR 2025
In this paper, we aim to build an adversarially robust zero-shot image classifier. We ground our work on CLIP, a vision-language pre-trained encoder model that can perform zero-shot classification by matching an image with text prompts ``a photo of a .''. Purification is the path we choose since it does not require adversarial training on specific attack types and thus can cope with any foreseen attacks. We then formulate purification risk as the KL divergence between the joint distributions of the purification process of denoising the adversarial samples and the attack process of adding perturbations to benign samples, through bidirectional Stochastic Differential Equations (SDEs). The final derived results inspire us to explore purification in the multi-modal latent space of CLIP. We propose two variants for our CLIPure approach: CLIPure-Diff which models the likelihood of images' latent vectors with the DiffusionPrior module in DaLLE-2 (modeling the generation process of CLIP's latent vectors), and CLIPure-Cos which models the likelihood with the cosine similarity between the embeddings of an image and ``a photo of a.''. As far as we know, CLIPure is the first purification method in multi-modal latent space and CLIPure-Cos is the first purification method that is not based on generative models, which substantially improves defense efficiency. We conducted extensive experiments on CIFAR-10, ImageNet, and 13 datasets that previous CLIP-based defense methods used for evaluating zero-shot classification robustness. Results show that CLIPure boosts the SOTA robustness by a large margin, e.g., from 71.7% to 91.1% on CIFAR10, from 59.6% to 72.6% on ImageNet, and 108% relative improvements of average robustness on the 13 datasets over previous SOTA. The code is available at https://github.com/TMLResearchGroup-CAS/CLIPure.
comment: accepted by ICLR 2025
♻ ☆ L-WISE: Boosting Human Visual Category Learning Through Model-Based Image Selection And Enhancement
The currently leading artificial neural network models of the visual ventral stream - which are derived from a combination of performance optimization and robustification methods - have demonstrated a remarkable degree of behavioral alignment with humans on visual categorization tasks. We show that image perturbations generated by these models can enhance the ability of humans to accurately report the ground truth class. Furthermore, we find that the same models can also be used out-of-the-box to predict the proportion of correct human responses to individual images, providing a simple, human-aligned estimator of the relative difficulty of each image. Motivated by these observations, we propose to augment visual learning in humans in a way that improves human categorization accuracy at test time. Our learning augmentation approach consists of (i) selecting images based on their model-estimated recognition difficulty, and (ii) applying image perturbations that aid recognition for novice learners. We find that combining these model-based strategies leads to categorization accuracy gains of 33-72% relative to control subjects without these interventions, on unmodified, randomly selected held-out test images. Beyond the accuracy gain, the training time for the augmented learning group was also shortened by 20-23%, despite both groups completing the same number of training trials. We demonstrate the efficacy of our approach in a fine-grained categorization task with natural images, as well as two tasks in clinically relevant image domains - histology and dermoscopy - where visual learning is notoriously challenging. To the best of our knowledge, our work is the first application of artificial neural networks to increase visual learning performance in humans by enhancing category-specific image features.
♻ ☆ Padding Tone: A Mechanistic Analysis of Padding Tokens in T2I Models NAACL 2025
Text-to-image (T2I) diffusion models rely on encoded prompts to guide the image generation process. Typically, these prompts are extended to a fixed length by adding padding tokens before text encoding. Despite being a default practice, the influence of padding tokens on the image generation process has not been investigated. In this work, we conduct the first in-depth analysis of the role padding tokens play in T2I models. We develop two causal techniques to analyze how information is encoded in the representation of tokens across different components of the T2I pipeline. Using these techniques, we investigate when and how padding tokens impact the image generation process. Our findings reveal three distinct scenarios: padding tokens may affect the model's output during text encoding, during the diffusion process, or be effectively ignored. Moreover, we identify key relationships between these scenarios and the model's architecture (cross or self-attention) and its training process (frozen or trained text encoder). These insights contribute to a deeper understanding of the mechanisms of padding tokens, potentially informing future model design and training practices in T2I systems.
comment: Published in: NAACL 2025. Project webpage: https://padding-tone.github.io/
♻ ☆ Pair-VPR: Place-Aware Pre-training and Contrastive Pair Classification for Visual Place Recognition with Vision Transformers
In this work we propose a novel joint training method for Visual Place Recognition (VPR), which simultaneously learns a global descriptor and a pair classifier for re-ranking. The pair classifier can predict whether a given pair of images are from the same place or not. The network only comprises Vision Transformer components for both the encoder and the pair classifier, and both components are trained using their respective class tokens. In existing VPR methods, typically the network is initialized using pre-trained weights from a generic image dataset such as ImageNet. In this work we propose an alternative pre-training strategy, by using Siamese Masked Image Modelling as a pre-training task. We propose a Place-aware image sampling procedure from a collection of large VPR datasets for pre-training our model, to learn visual features tuned specifically for VPR. By re-using the Mask Image Modelling encoder and decoder weights in the second stage of training, Pair-VPR can achieve state-of-the-art VPR performance across five benchmark datasets with a ViT-B encoder, along with further improvements in localization recall with larger encoders. The Pair-VPR website is: https://csiro-robotics.github.io/Pair-VPR.
♻ ☆ WalnutData: A UAV Remote Sensing Dataset of Green Walnuts and Model Evaluation
The UAV technology is gradually maturing and can provide extremely powerful support for smart agriculture and precise monitoring. Currently, there is no dataset related to green walnuts in the field of agricultural computer vision. Thus, in order to promote the algorithm design in the field of agricultural computer vision, we used UAV to collect remote-sensing data from 8 walnut sample plots. Considering that green walnuts are subject to various lighting conditions and occlusion, we constructed a large-scale dataset with a higher-granularity of target features - WalnutData. This dataset contains a total of 30,240 images and 706,208 instances, and there are 4 target categories: being illuminated by frontal light and unoccluded (A1), being backlit and unoccluded (A2), being illuminated by frontal light and occluded (B1), and being backlit and occluded (B2). Subsequently, we evaluated many mainstream algorithms on WalnutData and used these evaluation results as the baseline standard. The dataset and all evaluation results can be obtained at https://github.com/1wuming/WalnutData.
♻ ☆ High-Resolution Image Synthesis via Next-Token Prediction
Recently, autoregressive models have demonstrated remarkable performance in class-conditional image generation. However, the application of next-token prediction to high-resolution text-to-image generation remains largely unexplored. In this paper, we introduce \textbf{D-JEPA$\cdot$T2I}, an autoregressive model based on continuous tokens that incorporates innovations in both architecture and training strategy to generate high-quality, photorealistic images at arbitrary resolutions, up to 4K. Architecturally, we adopt the denoising joint embedding predictive architecture (D-JEPA) while leveraging a multimodal visual transformer to effectively integrate textual and visual features. Additionally, we introduce flow matching loss alongside the proposed Visual Rotary Positional Embedding (VoPE) to enable continuous resolution learning. In terms of training strategy, we propose a data feedback mechanism that dynamically adjusts the sampling procedure based on statistical analysis and an online learning critic model. This encourages the model to move beyond its comfort zone, reducing redundant training on well-mastered scenarios and compelling it to address more challenging cases with suboptimal generation quality. For the first time, we achieve state-of-the-art high-resolution image synthesis via next-token prediction.
comment: 31 pages
♻ ☆ ET-SEED: Efficient Trajectory-Level SE(3) Equivariant Diffusion Policy ICLR 2025
Imitation learning, e.g., diffusion policy, has been proven effective in various robotic manipulation tasks. However, extensive demonstrations are required for policy robustness and generalization. To reduce the demonstration reliance, we leverage spatial symmetry and propose ET-SEED, an efficient trajectory-level SE(3) equivariant diffusion model for generating action sequences in complex robot manipulation tasks. Further, previous equivariant diffusion models require the per-step equivariance in the Markov process, making it difficult to learn policy under such strong constraints. We theoretically extend equivariant Markov kernels and simplify the condition of equivariant diffusion process, thereby significantly improving training efficiency for trajectory-level SE(3) equivariant diffusion policy in an end-to-end manner. We evaluate ET-SEED on representative robotic manipulation tasks, involving rigid body, articulated and deformable object. Experiments demonstrate superior data efficiency and manipulation proficiency of our proposed method, as well as its ability to generalize to unseen configurations with only a few demonstrations. Website: https://et-seed.github.io/
comment: Accept to ICLR 2025
♻ ☆ Autoregressive Video Generation without Vector Quantization ICLR 2025
This paper presents a novel approach that enables autoregressive video generation with high efficiency. We propose to reformulate the video generation problem as a non-quantized autoregressive modeling of temporal frame-by-frame prediction and spatial set-by-set prediction. Unlike raster-scan prediction in prior autoregressive models or joint distribution modeling of fixed-length tokens in diffusion models, our approach maintains the causal property of GPT-style models for flexible in-context capabilities, while leveraging bidirectional modeling within individual frames for efficiency. With the proposed approach, we train a novel video autoregressive model without vector quantization, termed NOVA. Our results demonstrate that NOVA surpasses prior autoregressive video models in data efficiency, inference speed, visual fidelity, and video fluency, even with a much smaller model capacity, i.e., 0.6B parameters. NOVA also outperforms state-of-the-art image diffusion models in text-to-image generation tasks, with a significantly lower training cost. Additionally, NOVA generalizes well across extended video durations and enables diverse zero-shot applications in one unified model. Code and models are publicly available at https://github.com/baaivision/NOVA.
comment: Accepted to ICLR 2025. Project page at https://github.com/baaivision/NOVA
♻ ☆ SEED-X: Multimodal Models with Unified Multi-granularity Comprehension and Generation
The rapid evolution of multimodal foundation model has demonstrated significant progresses in vision-language understanding and generation, e.g., our previous work SEED-LLaMA. However, there remains a gap between its capability and the real-world applicability, primarily due to the model's limited capacity to effectively respond to various user instructions and interact with diverse visual data. In this work, we focus on bridging this gap through integrating two enhanced features: (1) comprehending images of arbitrary sizes and ratios, and (2) enabling multi-granularity image generation. We present a unified and versatile foundation model, namely, SEED-X, which is able to model multi-granularity visual semantics for comprehension and generation tasks. Besides the competitive results on public benchmarks, SEED-X demonstrates its effectiveness in handling real-world applications across various domains after instruction tuning. We hope that our work will inspire future research into what can be achieved by versatile multimodal foundation models in real-world applications. The models, codes, and datasets are released in https://github.com/AILab-CVC/SEED-X.
comment: We added benchmark results (without updating models) and ablation study in this version. Project released at: https://github.com/AILab-CVC/SEED-X
♻ ☆ Predictive Uncertainty Quantification for Bird's Eye View Segmentation: A Benchmark and Novel Loss Function ICLR 2025
The fusion of raw sensor data to create a Bird's Eye View (BEV) representation is critical for autonomous vehicle planning and control. Despite the growing interest in using deep learning models for BEV semantic segmentation, anticipating segmentation errors and enhancing the explainability of these models remain underexplored. This paper introduces a comprehensive benchmark for predictive uncertainty quantification in BEV segmentation, evaluating multiple uncertainty quantification methods across three popular datasets with three representative network architectures. Our study focuses on the effectiveness of quantified uncertainty in detecting misclassified and out-of-distribution (OOD) pixels while also improving model calibration. Through empirical analysis, we uncover challenges in existing uncertainty quantification methods and demonstrate the potential of evidential deep learning techniques, which capture both aleatoric and epistemic uncertainty. To address these challenges, we propose a novel loss function, Uncertainty-Focal-Cross-Entropy (UFCE), specifically designed for highly imbalanced data, along with a simple uncertainty-scaling regularization term that improves both uncertainty quantification and model calibration for BEV segmentation.
comment: ICLR 2025
♻ ☆ LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding ICLR 2025
Auto-Regressive (AR) models have recently gained prominence in image generation, often matching or even surpassing the performance of diffusion models. However, one major limitation of AR models is their sequential nature, which processes tokens one at a time, slowing down generation compared to models like GANs or diffusion-based methods that operate more efficiently. While speculative decoding has proven effective for accelerating LLMs by generating multiple tokens in a single forward, its application in visual AR models remains largely unexplored. In this work, we identify a challenge in this setting, which we term \textit{token selection ambiguity}, wherein visual AR models frequently assign uniformly low probabilities to tokens, hampering the performance of speculative decoding. To overcome this challenge, we propose a relaxed acceptance condition referred to as LANTERN that leverages the interchangeability of tokens in latent space. This relaxation restores the effectiveness of speculative decoding in visual AR models by enabling more flexible use of candidate tokens that would otherwise be prematurely rejected. Furthermore, by incorporating a total variation distance bound, we ensure that these speed gains are achieved without significantly compromising image quality or semantic coherence. Experimental results demonstrate the efficacy of our method in providing a substantial speed-up over speculative decoding. In specific, compared to a na\"ive application of the state-of-the-art speculative decoding, LANTERN increases speed-ups by $\mathbf{1.75}\times$ and $\mathbf{1.82}\times$, as compared to greedy decoding and random sampling, respectively, when applied to LlamaGen, a contemporary visual AR model. The code is publicly available at https://github.com/jadohu/LANTERN.
comment: 30 pages, 13 figures, Accepted to ICLR 2025 (poster)
♻ ☆ InterMask: 3D Human Interaction Generation via Collaborative Masked Modeling
Generating realistic 3D human-human interactions from textual descriptions remains a challenging task. Existing approaches, typically based on diffusion models, often produce results lacking realism and fidelity. In this work, we introduce InterMask, a novel framework for generating human interactions using collaborative masked modeling in discrete space. InterMask first employs a VQ-VAE to transform each motion sequence into a 2D discrete motion token map. Unlike traditional 1D VQ token maps, it better preserves fine-grained spatio-temporal details and promotes spatial awareness within each token. Building on this representation, InterMask utilizes a generative masked modeling framework to collaboratively model the tokens of two interacting individuals. This is achieved by employing a transformer architecture specifically designed to capture complex spatio-temporal inter-dependencies. During training, it randomly masks the motion tokens of both individuals and learns to predict them. For inference, starting from fully masked sequences, it progressively fills in the tokens for both individuals. With its enhanced motion representation, dedicated architecture, and effective learning strategy, InterMask achieves state-of-the-art results, producing high-fidelity and diverse human interactions. It outperforms previous methods, achieving an FID of $5.154$ (vs $5.535$ of in2IN) on the InterHuman dataset and $0.399$ (vs $5.207$ of InterGen) on the InterX dataset. Additionally, InterMask seamlessly supports reaction generation without the need for model redesign or fine-tuning.
comment: Project webpage: https://gohar-malik.github.io/intermask
♻ ☆ MoCoLSK: Modality Conditioned High-Resolution Downscaling for Land Surface Temperature
Land Surface Temperature (LST) is a critical parameter for environmental studies, but directly obtaining high spatial resolution LST data remains challenging due to the spatio-temporal trade-off in satellite remote sensing. Guided LST downscaling has emerged as an alternative solution to overcome these limitations, but current methods often neglect spatial non-stationarity, and there is a lack of an open-source ecosystem for deep learning methods. In this paper, we propose the Modality-Conditional Large Selective Kernel (MoCoLSK) Network, a novel architecture that dynamically fuses multi-modal data through modality-conditioned projections. MoCoLSK achieves a confluence of dynamic receptive field adjustment and multi-modal feature fusion, leading to enhanced LST prediction accuracy. Furthermore, we establish the GrokLST project, a comprehensive open-source ecosystem featuring the GrokLST dataset, a high-resolution benchmark, and the GrokLST toolkit, an open-source PyTorch-based toolkit encapsulating MoCoLSK alongside 40+ state-of-the-art approaches. Extensive experimental results validate MoCoLSK's effectiveness in capturing complex dependencies and subtle variations within multispectral data, outperforming existing methods in LST downscaling. Our code, dataset, and toolkit are available at https://github.com/GrokCV/GrokLST.
comment: Accepted by IEEE TGRS
♻ ☆ TEASER: Token Enhanced Spatial Modeling for Expressions Reconstruction ICLR 2025
3D facial reconstruction from a single in-the-wild image is a crucial task in human-centered computer vision tasks. While existing methods can recover accurate facial shapes, there remains significant space for improvement in fine-grained expression capture. Current approaches struggle with irregular mouth shapes, exaggerated expressions, and asymmetrical facial movements. We present TEASER (Token EnhAnced Spatial modeling for Expressions Reconstruction), which addresses these challenges and enhances 3D facial geometry performance. TEASER tackles two main limitations of existing methods: insufficient photometric loss for self-reconstruction and inaccurate localization of subtle expressions. We introduce a multi-scale tokenizer to extract facial appearance information. Combined with a neural renderer, these tokens provide precise geometric guidance for expression reconstruction. Furthermore, TEASER incorporates a pose-dependent landmark loss to further improve geometric performances. Our approach not only significantly enhances expression reconstruction quality but also offers interpretable tokens suitable for various downstream applications, such as photorealistic facial video driving, expression transfer, and identity swapping. Quantitative and qualitative experimental results across multiple datasets demonstrate that TEASER achieves state-of-the-art performance in precise expression reconstruction.
comment: Accepted by ICLR 2025, code and demos are available at https://tinyurl.com/TEASER-project
♻ ☆ Improving vision-language alignment with graph spiking hybrid Networks
To bridge the semantic gap between vision and language (VL), it is necessary to develop a good alignment strategy, which includes handling semantic diversity, abstract representation of visual information, and generalization ability of models. Recent works use detector-based bounding boxes or patches with regular partitions to represent visual semantics. While current paradigms have made strides, they are still insufficient for fully capturing the nuanced contextual relations among various objects. This paper proposes a comprehensive visual semantic representation module, necessitating the utilization of panoptic segmentation to generate coherent fine-grained semantic features. Furthermore, we propose a novel Graph Spiking Hybrid Network (GSHN) that integrates the complementary advantages of Spiking Neural Networks (SNNs) and Graph Attention Networks (GATs) to encode visual semantic information. Intriguingly, the model not only encodes the discrete and continuous latent variables of instances but also adeptly captures both local and global contextual features, thereby significantly enhancing the richness and diversity of semantic representations. Leveraging the spatiotemporal properties inherent in SNNs, we employ contrastive learning (CL) to enhance the similarity-based representation of embeddings. This strategy alleviates the computational overhead of the model and enriches meaningful visual representations by constructing positive and negative sample pairs. We design an innovative pre-training method, Spiked Text Learning (STL), which uses text features to improve the encoding ability of discrete semantics. Experiments show that the proposed GSHN exhibits promising results on multiple VL downstream tasks.
♻ ☆ Improving Long-Text Alignment for Text-to-Image Diffusion Models
The rapid advancement of text-to-image (T2I) diffusion models has enabled them to generate unprecedented results from given texts. However, as text inputs become longer, existing encoding methods like CLIP face limitations, and aligning the generated images with long texts becomes challenging. To tackle these issues, we propose LongAlign, which includes a segment-level encoding method for processing long texts and a decomposed preference optimization method for effective alignment training. For segment-level encoding, long texts are divided into multiple segments and processed separately. This method overcomes the maximum input length limits of pretrained encoding models. For preference optimization, we provide decomposed CLIP-based preference models to fine-tune diffusion models. Specifically, to utilize CLIP-based preference models for T2I alignment, we delve into their scoring mechanisms and find that the preference scores can be decomposed into two components: a text-relevant part that measures T2I alignment and a text-irrelevant part that assesses other visual aspects of human preference. Additionally, we find that the text-irrelevant part contributes to a common overfitting problem during fine-tuning. To address this, we propose a reweighting strategy that assigns different weights to these two components, thereby reducing overfitting and enhancing alignment. After fine-tuning $512 \times 512$ Stable Diffusion (SD) v1.5 for about 20 hours using our method, the fine-tuned SD outperforms stronger foundation models in T2I alignment, such as PixArt-$\alpha$ and Kandinsky v2.2. The code is available at https://github.com/luping-liu/LongAlign.
♻ ☆ SCC-YOLO: An Improved Object Detector for Assisting in Brain Tumor Diagnosis
Brain tumors can lead to neurological dysfunction, cognitive and psychological changes, increased intracranial pressure, and seizures, posing significant risks to health. The You Only Look Once (YOLO) series has shown superior accuracy in medical imaging object detection. This paper presents a novel SCC-YOLO architecture that integrates the SCConv module into YOLOv9. The SCConv module optimizes convolutional efficiency by reducing spatial and channel redundancy, enhancing image feature learning. We examine the effects of different attention mechanisms with YOLOv9 for brain tumor detection using the Br35H dataset and our custom dataset (Brain_Tumor_Dataset). Results indicate that SCC-YOLO improved mAP50 by 0.3% on the Br35H dataset and by 0.5% on our custom dataset compared to YOLOv9. SCC-YOLO achieves state-of-the-art performance in brain tumor detection.
♻ ☆ IRisPath: Enhancing Costmap for Off-Road Navigation with Robust IR-RGB Fusion for Improved Day and Night Traversability
Autonomous off-road navigation is required for applications in agriculture, construction, search and rescue and defence. Traditional on-road autonomous methods struggle with dynamic terrains, leading to poor vehicle control in off-road conditions. Recent deep-learning models have used perception sensors along with kinesthetic feedback for navigation on such terrains. However, this approach has out-of-domain uncertainty. Factors like change in time of day and weather impacts the performance of the model. We propose a multi modal fusion network "IRisPath" capable of using Thermal and RGB images to provide robustness against dynamic weather and light conditions. To aid further works in this domain, we also open-source a day-night dataset with Thermal and RGB images along with pseudo-labels for traversability. In order to co-register for fusion model we also develop a novel method for targetless extrinsic calibration of Thermal, LiDAR and RGB cameras with translation accuracy of +/-1.7cm and rotation accuracy of +/-0.827degrees.
♻ ☆ End-to-End Augmentation Hyperparameter Tuning for Self-Supervised Anomaly Detection
Self-supervised learning (SSL) has emerged as a promising paradigm that presents supervisory signals to real-world problems, bypassing the extensive cost of manual labeling. Consequently, self-supervised anomaly detection (SSAD) has seen a recent surge of interest, since SSL is especially attractive for unsupervised tasks. However, recent works have reported that the choice of a data augmentation function has significant impact on the accuracy of SSAD, posing augmentation search as an essential but nontrivial problem with the lack of labeled validation data. In this paper, we introduce ST-SSAD, the first systematic approach for rigorous augmentation tuning on SSAD. To this end, our work presents two key contributions. The first is a new unsupervised validation loss that quantifies the alignment between augmented training data and unlabeled validation data. The second is new differentiable augmentation functions, allowing data augmentation hyperparameter(s) to be tuned in an end-to-end manner. Experiments on two testbeds with semantic class anomalies and subtle industrial defects show that ST-SSAD gives significant performance gains over existing works.
♻ ☆ AuroraCap: Efficient, Performant Video Detailed Captioning and a New Benchmark ICLR 2025
Video detailed captioning is a key task which aims to generate comprehensive and coherent textual descriptions of video content, benefiting both video understanding and generation. In this paper, we propose AuroraCap, a video captioner based on a large multimodal model. We follow the simplest architecture design without additional parameters for temporal modeling. To address the overhead caused by lengthy video sequences, we implement the token merging strategy, reducing the number of input visual tokens. Surprisingly, we found that this strategy results in little performance loss. AuroraCap shows superior performance on various video and image captioning benchmarks, for example, obtaining a CIDEr of 88.9 on Flickr30k, beating GPT-4V (55.3) and Gemini-1.5 Pro (82.2). However, existing video caption benchmarks only include simple descriptions, consisting of a few dozen words, which limits research in this field. Therefore, we develop VDC, a video detailed captioning benchmark with over one thousand carefully annotated structured captions. In addition, we propose a new LLM-assisted metric VDCscore for bettering evaluation, which adopts a divide-and-conquer strategy to transform long caption evaluation into multiple short question-answer pairs. With the help of human Elo ranking, our experiments show that this benchmark better correlates with human judgments of video detailed captioning quality.
comment: Accepted to ICLR 2025. Code, docs, weight, benchmark and training data are all avaliable at https://rese1f.github.io/aurora-web/
♻ ☆ EMT: A Visual Multi-Task Benchmark Dataset for Autonomous Driving in the Arab Gulf Region
This paper introduces the Emirates Multi-Task (EMT) dataset - the first publicly available dataset for autonomous driving collected in the Arab Gulf region. The EMT dataset captures the unique road topology, high traffic congestion, and distinctive characteristics of the Gulf region, including variations in pedestrian clothing and weather conditions. It contains over 30,000 frames from a dash-camera perspective, along with 570,000 annotated bounding boxes, covering approximately 150 kilometers of driving routes. The EMT dataset supports three primary tasks: tracking, trajectory forecasting and intention prediction. Each benchmark dataset is complemented with corresponding evaluations: (1) multi-agent tracking experiments, focusing on multi-class scenarios and occlusion handling; (2) trajectory forecasting evaluation using deep sequential and interaction-aware models; and (3) intention benchmark experiments conducted for predicting agents intentions from observed trajectories. The dataset is publicly available at avlab.io/emt-dataset, and pre-processing scripts along with evaluation models can be accessed at github.com/AV-Lab/emt-dataset.
comment: 19 pages, 6 figures
♻ ☆ Measuring Anxiety Levels with Head Motion Patterns in Severe Depression Population
Depression and anxiety are prevalent mental health disorders that frequently cooccur, with anxiety significantly influencing both the manifestation and treatment of depression. An accurate assessment of anxiety levels in individuals with depression is crucial to develop effective and personalized treatment plans. This study proposes a new noninvasive method for quantifying anxiety severity by analyzing head movements -- specifically speed, acceleration, and angular displacement -- during video-recorded interviews with patients suffering from severe depression. Using data from a new CALYPSO Depression Dataset, we extracted head motion characteristics and applied regression analysis to predict clinically evaluated anxiety levels. Our results demonstrate a high level of precision, achieving a mean absolute error (MAE) of 0.35 in predicting the severity of psychological anxiety based on head movement patterns. This indicates that our approach can enhance the understanding of anxiety's role in depression and assist psychiatrists in refining treatment strategies for individuals.
comment: 19th IEEE International Conference on Automatic Face and Gesture Recognition (FG), 2025
♻ ☆ Few-Class Arena: A Benchmark for Efficient Selection of Vision Models and Dataset Difficulty Measurement
We propose Few-Class Arena (FCA), as a unified benchmark with focus on testing efficient image classification models for few classes. A wide variety of benchmark datasets with many classes (80-1000) have been created to assist Computer Vision architectural evolution. An increasing number of vision models are evaluated with these many-class datasets. However, real-world applications often involve substantially fewer classes of interest (2-10). This gap between many and few classes makes it difficult to predict performance of the few-class applications using models trained on the available many-class datasets. To date, little has been offered to evaluate models in this Few-Class Regime. We conduct a systematic evaluation of the ResNet family trained on ImageNet subsets from 2 to 1000 classes, and test a wide spectrum of Convolutional Neural Networks and Transformer architectures over ten datasets by using our newly proposed FCA tool. Furthermore, to aid an up-front assessment of dataset difficulty and a more efficient selection of models, we incorporate a difficulty measure as a function of class similarity. FCA offers a new tool for efficient machine learning in the Few-Class Regime, with goals ranging from a new efficient class similarity proposal, to lightweight model architecture design, to a new scaling law. FCA is user-friendly and can be easily extended to new models and datasets, facilitating future research work. Our benchmark is available at https://github.com/bryanbocao/fca.
comment: 10 pages, 32 pages including References and Appendix, 19 figures, 8 tables
♻ ☆ Modeling and Analysis of Spatial and Temporal Land Clutter Statistics in SAR Imaging Based on MSTAR Data
The statistical analysis of land clutter for Synthetic Aperture Radar (SAR) imaging has become an increasingly important subject for research and investigation. It is also absolutely necessary for designing robust algorithms capable of performing the task of target detection in the background clutter. Any attempt to extract the energy of the desired targets from the land clutter requires complete knowledge of the statistical properties of the background clutter. In this paper, the spatial as well as the temporal characteristics of the land clutter are studied. Since the data for each image has been collected based on a different aspect angle; therefore, the temporal analysis contains variation in the aspect angle. Consequently, the temporal analysis includes the characteristics of the radar cross section with respect to the aspect angle based on which the data has been collected. In order to perform the statistical analysis, several well-known and relevant distributions, namely, Weibull, Log-normal, Gamma, and Rayleigh are considered as prime candidates to model the land clutter. The goodness-of-fit test is based on the Kullback-Leibler (KL) Divergence metric. The detailed analysis presented in this paper demonstrates that the Weibull distribution is a more accurate fit for the temporal-aspect-angle statistical analysis while the Rayleigh distribution models the spatial characteristics of the background clutter with higher accuracy. Finally, based on the aforementioned statistical analyses and by utilizing the Constant False Alarm Rate (CFAR) algorithm, we perform target detection in land clutter. The overall verification of the analysis is performed by exploiting the Moving and Stationary Target Acquisition and Recognition (MSTAR) data-set, which has been collected in spotlight mode at X-band, and the results are presented.
comment: arXiv admin note: substantial text overlap with arXiv:2409.02155
♻ ☆ Snuffy: Efficient Whole Slide Image Classifier ECCV 2024
Whole Slide Image (WSI) classification with multiple instance learning (MIL) in digital pathology faces significant computational challenges. Current methods mostly rely on extensive self-supervised learning (SSL) for satisfactory performance, requiring long training periods and considerable computational resources. At the same time, no pre-training affects performance due to domain shifts from natural images to WSIs. We introduce Snuffy architecture, a novel MIL-pooling method based on sparse transformers that mitigates performance loss with limited pre-training and enables continual few-shot pre-training as a competitive option. Our sparsity pattern is tailored for pathology and is theoretically proven to be a universal approximator with the tightest probabilistic sharp bound on the number of layers for sparse transformers, to date. We demonstrate Snuffy's effectiveness on CAMELYON16 and TCGA Lung cancer datasets, achieving superior WSI and patch-level accuracies. The code is available on https://github.com/jafarinia/snuffy.
comment: Accepted for ECCV 2024
♻ ☆ DynRefer: Delving into Region-level Multimodal Tasks via Dynamic Resolution CVPR 2025
One fundamental task of multimodal models is to translate referred image regions to human preferred language descriptions. Existing methods, however, ignore the resolution adaptability needs of different tasks, which hinders them to find out precise language descriptions. In this study, we propose a DynRefer approach, to pursue high-accuracy region-level referring through mimicking the resolution adaptability of human visual cognition. During training, DynRefer stochastically aligns language descriptions of multimodal tasks with images of multiple resolutions, which are constructed by nesting a set of random views around the referred region. During inference, DynRefer performs selectively multimodal referring by sampling proper region representations for tasks from the nested views based on image and task priors. This allows the visual information for referring to better match human preferences, thereby improving the representational adaptability of region-level multimodal models. Experiments show that DynRefer brings mutual improvement upon broad tasks including region-level captioning, open-vocabulary region recognition and attribute detection. Furthermore, DynRefer achieves state-of-the-art results on multiple region-level multimodal tasks using a single model. Code is available at https://github.com/callsys/DynRefer.
comment: Accepted in CVPR 2025. Code is available at https://github.com/callsys/DynRefer
♻ ☆ Neural Finite-State Machines for Surgical Phase Recognition
Surgical phase recognition (SPR) is crucial for applications in workflow optimization, performance evaluation, and real-time intervention guidance. However, current deep learning models often struggle with fragmented predictions, failing to capture the sequential nature of surgical workflows. We propose the Neural Finite-State Machine (NFSM), a novel approach that enforces temporal coherence by integrating classical state-transition priors with modern neural networks. NFSM leverages learnable global state embeddings as unique phase identifiers and dynamic transition tables to model phase-to-phase progressions. Additionally, a future phase forecasting mechanism employs repeated frame padding to anticipate upcoming transitions. Implemented as a plug-and-play module, NFSM can be integrated into existing SPR pipelines without changing their core architectures. We demonstrate state-of-the-art performance across multiple benchmarks, including a significant improvement on the BernBypass70 dataset - raising video-level accuracy by 0.9 points and phase-level precision, recall, F1-score, and mAP by 3.8, 3.1, 3.3, and 4.1, respectively. Ablation studies confirm each component's effectiveness and the module's adaptability to various architectures. By unifying finite-state principles with deep learning, NFSM offers a robust path toward consistent, long-term surgical video analysis.
♻ ☆ Refinement Module based on Parse Graph of Feature Map for Human Pose Estimation
Parse graphs of the human body can be obtained in the human brain to help humans complete the human Pose Estimation better (HPE). It contains a hierarchical structure, like a tree structure, and context relations among nodes. To equip models with such capabilities, many researchers predefine the parse graph of body structure to design HPE frameworks. However, these frameworks struggle to adapt to instances that deviate from the predefined parse graph and are often parameter-heavy. Unlike them, we view the feature map holistically, much like the human body. It can be optimized using parse graphs, where each node's feature is an implicit expression rather than a fixed one. This allows it to adapt to more instances, unconstrained by rigid structural features. In this paper, we design the Refinement Module based on the Parse Graph of feature map (RMPG), which includes two stages: top-down decomposition and bottom-up combination. In the first stage, the feature map is decomposed into multiple sub-feature maps along the channel. In the second stage, the context relations of sub-feature maps are calculated to obtain their respective context information and the sub-feature maps with context information are concatenated along channels to obtain the refined feature map. Additionally, we design a hierarchical network with fewer parameters using multiple RMPG modules to model the context relations and hierarchies in the parse graph of body structure for HPE, some of which are supervised to obtain context relations among body parts. Our network achieves excellent results on multiple mainstream human pose datasets. More importantly, the effectiveness of RMPG is proven on different methods. The code of RMPG will be open.
♻ ☆ Towards Robust Algorithms for Surgical Phase Recognition via Digital Twin Representation
Surgical phase recognition (SPR) is an integral component of surgical data science, enabling high-level surgical analysis. End-to-end trained neural networks that predict surgical phase directly from videos have shown excellent performance on benchmarks. However, these models struggle with robustness due to non-causal associations in the training set. Our goal is to improve model robustness to variations in the surgical videos by leveraging the digital twin (DT) paradigm -- an intermediary layer to separate high-level analysis (SPR) from low-level processing. As a proof of concept, we present a DT representation-based framework for SPR from videos. The framework employs vision foundation models with reliable low-level scene understanding to craft DT representation. We embed the DT representation in place of raw video inputs in the state-of-the-art SPR model. The framework is trained on the Cholec80 dataset and evaluated on out-of-distribution (OOD) and corrupted test samples. Contrary to the vulnerability of the baseline model, our framework demonstrates strong robustness on both OOD and corrupted samples, with a video-level accuracy of 80.3 on a highly corrupted Cholec80 test set, 67.9 on the challenging CRCD dataset, and 99.8 on an internal robotic surgery dataset, outperforming the baseline by 3.9, 16.8, and 90.9 respectively. We also find that using DT representation as an augmentation to the raw input can significantly improve model robustness. Our findings lend support to the thesis that DT representations are effective in enhancing model robustness. Future work will seek to improve the feature informativeness and incorporate interpretability for a more comprehensive framework.
♻ ☆ SoK: Systematization and Benchmarking of Deepfake Detectors in a Unified Framework EuroS&P '25
Deepfakes have rapidly emerged as a serious threat to society due to their ease of creation and dissemination, triggering the accelerated development of detection technologies. However, many existing detectors rely on labgenerated datasets for validation, which may not prepare them for novel, real-world deepfakes. This paper extensively reviews and analyzes state-of-the-art deepfake detectors, evaluating them against several critical criteria. These criteria categorize detectors into 4 high-level groups and 13 finegrained sub-groups, aligned with a unified conceptual framework we propose. This classification offers practical insights into the factors affecting detector efficacy. We evaluate the generalizability of 16 leading detectors across comprehensive attack scenarios, including black-box, white-box, and graybox settings. Our systematized analysis and experiments provide a deeper understanding of deepfake detectors and their generalizability, paving the way for future research and the development of more proactive defenses against deepfakes.
comment: 20 pages, 6 figures, 7 table, Accepted at IEEE European Symposium on security and privacy 2025 (EuroS&P '25)
♻ ☆ Image Watermarks are Removable Using Controllable Regeneration from Clean Noise ICLR2025
Image watermark techniques provide an effective way to assert ownership, deter misuse, and trace content sources, which has become increasingly essential in the era of large generative models. A critical attribute of watermark techniques is their robustness against various manipulations. In this paper, we introduce a watermark removal approach capable of effectively nullifying state-of-the-art watermarking techniques. Our primary insight involves regenerating the watermarked image starting from a clean Gaussian noise via a controllable diffusion model, utilizing the extracted semantic and spatial features from the watermarked image. The semantic control adapter and the spatial control network are specifically trained to control the denoising process towards ensuring image quality and enhancing consistency between the cleaned image and the original watermarked image. To achieve a smooth trade-off between watermark removal performance and image consistency, we further propose an adjustable and controllable regeneration scheme. This scheme adds varying numbers of noise steps to the latent representation of the watermarked image, followed by a controlled denoising process starting from this noisy latent representation. As the number of noise steps increases, the latent representation progressively approaches clean Gaussian noise, facilitating the desired trade-off. We apply our watermark removal methods across various watermarking techniques, and the results demonstrate that our methods offer superior visual consistency/quality and enhanced watermark removal performance compared to existing regeneration approaches. Our code is available at https://github.com/yepengliu/CtrlRegen.
comment: ICLR2025
♻ ☆ Adaptive Neural Networks for Intelligent Data-Driven Development
Advances in machine learning methods for computer vision tasks have led to their consideration for safety-critical applications like autonomous driving. However, effectively integrating these methods into the automotive development lifecycle remains challenging. Since the performance of machine learning algorithms relies heavily on the training data provided, the data and model development lifecycle play a key role in successfully integrating these components into the product development lifecycle. Existing models frequently encounter difficulties recognizing or adapting to novel instances not present in the original training dataset. This poses a significant risk for reliable deployment in dynamic environments. To address this challenge, we propose an adaptive neural network architecture and an iterative development framework that enables users to efficiently incorporate previously unknown objects into the current perception system. Our approach builds on continuous learning, emphasizing the necessity of dynamic updates to reflect real-world deployment conditions. Specifically, we introduce a pipeline with three key components: (1) a scalable network extension strategy to integrate new classes while preserving existing performance, (2) a dynamic OoD detection component that requires no additional retraining for newly added classes, and (3) a retrieval-based data augmentation process tailored for safety-critical deployments. The integration of these components establishes a pragmatic and adaptive pipeline for the continuous evolution of perception systems in the context of autonomous driving.
comment: 8 pages, 3 figures, and 3 tables
♻ ☆ Segment-Level Road Obstacle Detection Using Visual Foundation Model Priors and Likelihood Ratios
Detecting road obstacles is essential for autonomous vehicles to navigate dynamic and complex traffic environments safely. Current road obstacle detection methods typically assign a score to each pixel and apply a threshold to generate final predictions. However, selecting an appropriate threshold is challenging, and the per-pixel classification approach often leads to fragmented predictions with numerous false positives. In this work, we propose a novel method that leverages segment-level features from visual foundation models and likelihood ratios to predict road obstacles directly. By focusing on segments rather than individual pixels, our approach enhances detection accuracy, reduces false positives, and offers increased robustness to scene variability. We benchmark our approach against existing methods on the RoadObstacle and LostAndFound datasets, achieving state-of-the-art performance without needing a predefined threshold.
comment: 10 pages, 4 figures, and 1 table, to be published in VISAPP 2025
♻ ☆ VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents
Retrieval-augmented generation (RAG) is an effective technique that enables large language models (LLMs) to utilize external knowledge sources for generation. However, current RAG systems are solely based on text, rendering it impossible to utilize vision information like layout and images that play crucial roles in real-world multi-modality documents. In this paper, we introduce VisRAG, which tackles this issue by establishing a vision-language model (VLM)-based RAG pipeline. In this pipeline, instead of first parsing the document to obtain text, the document is directly embedded using a VLM as an image and then retrieved to enhance the generation of a VLM. Compared to traditional text-based RAG, VisRAG maximizes the retention and utilization of the data information in the original documents, eliminating the information loss introduced during the parsing process. We collect both open-source and synthetic data to train the retriever in VisRAG and explore a variety of generation methods. Experiments demonstrate that VisRAG outperforms traditional RAG in both the retrieval and generation stages, achieving a 20--40% end-to-end performance gain over traditional text-based RAG pipeline. Further analysis reveals that VisRAG is efficient in utilizing training data and demonstrates strong generalization capability, positioning it as a promising solution for RAG on multi-modality documents. Our code and data are available at https://github.com/openbmb/visrag.
♻ ☆ Score Forgetting Distillation: A Swift, Data-Free Method for Machine Unlearning in Diffusion Models ICLR 2025
The machine learning community is increasingly recognizing the importance of fostering trust and safety in modern generative AI (GenAI) models. We posit machine unlearning (MU) as a crucial foundation for developing safe, secure, and trustworthy GenAI models. Traditional MU methods often rely on stringent assumptions and require access to real data. This paper introduces Score Forgetting Distillation (SFD), an innovative MU approach that promotes the forgetting of undesirable information in diffusion models by aligning the conditional scores of "unsafe" classes or concepts with those of "safe" ones. To eliminate the need for real data, our SFD framework incorporates a score-based MU loss into the score distillation objective of a pretrained diffusion model. This serves as a regularization term that preserves desired generation capabilities while enabling the production of synthetic data through a one-step generator. Our experiments on pretrained label-conditional and text-to-image diffusion models demonstrate that our method effectively accelerates the forgetting of target classes or concepts during generation, while preserving the quality of other classes or concepts. This unlearned and distilled diffusion not only pioneers a novel concept in MU but also accelerates the generation speed of diffusion models. Our experiments and studies on a range of diffusion models and datasets confirm that our approach is generalizable, effective, and advantageous for MU in diffusion models. Code is available at https://github.com/tqch/score-forgetting-distillation. ($\textbf{Warning:}$ This paper contains sexually explicit imagery, discussions of pornography, racially-charged terminology, and other content that some readers may find disturbing, distressing, and/or offensive.)
comment: ICLR 2025
♻ ☆ GP-GS: Gaussian Processes for Enhanced Gaussian Splatting
3D Gaussian Splatting has emerged as an efficient photorealistic novel view synthesis method. However, its reliance on sparse Structure-from-Motion (SfM) point clouds consistently compromises the scene reconstruction quality. To address these limitations, this paper proposes a novel 3D reconstruction framework Gaussian Processes Gaussian Splatting (GP-GS), where a multi-output Gaussian Process model is developed to achieve adaptive and uncertainty-guided densification of sparse SfM point clouds. Specifically, we propose a dynamic sampling and filtering pipeline that adaptively expands the SfM point clouds by leveraging GP-based predictions to infer new candidate points from the input 2D pixels and depth maps. The pipeline utilizes uncertainty estimates to guide the pruning of high-variance predictions, ensuring geometric consistency and enabling the generation of dense point clouds. The densified point clouds provide high-quality initial 3D Gaussians to enhance reconstruction performance. Extensive experiments conducted on synthetic and real-world datasets across various scales validate the effectiveness and practicality of the proposed framework.
comment: 14 pages,11 figures
♻ ☆ All Seeds Are Not Equal: Enhancing Compositional Text-to-Image Generation with Reliable Random Seeds
Text-to-image diffusion models have demonstrated remarkable capability in generating realistic images from arbitrary text prompts. However, they often produce inconsistent results for compositional prompts such as "two dogs" or "a penguin on the right of a bowl". Understanding these inconsistencies is crucial for reliable image generation. In this paper, we highlight the significant role of initial noise in these inconsistencies, where certain noise patterns are more reliable for compositional prompts than others. Our analyses reveal that different initial random seeds tend to guide the model to place objects in distinct image areas, potentially adhering to specific patterns of camera angles and image composition associated with the seed. To improve the model's compositional ability, we propose a method for mining these reliable cases, resulting in a curated training set of generated images without requiring any manual annotation. By fine-tuning text-to-image models on these generated images, we significantly enhance their compositional capabilities. For numerical composition, we observe relative increases of 29.3% and 19.5% for Stable Diffusion and PixArt-{\alpha}, respectively. Spatial composition sees even larger gains, with 60.7% for Stable Diffusion and 21.1% for PixArt-{\alpha}.
Information Retrieval 9
☆ Can We Find the Code? An Empirical Study of Google Scholar's Code Retrieval
Academic codes associated with research papers are valuable resources for scholars. In specialized fields outside computer science, code availability is often limited, making effective code retrieval essential. Google Scholar is a crucial academic search tool. If a code published in the paper is not retrievable via Google Scholar, its accessibility and impact are significantly reduced. This study takes the term "accelerated degradation" combined with "reliability" as an example, and finds that, for papers published by Elsevier, only GitHub links included in abstracts are comprehensively retrieved by Google Scholar. When such links appear within the main body of a paper, even in the "Data Availability" section, they may be ignored and become unsearchable. These findings highlight the importance of strategically placing GitHub links in abstracts to enhance code discoverability on Google Scholar.
☆ A Semantic Search Pipeline for Causality-driven Adhoc Information Retrieval
We present a unsupervised semantic search pipeline for the Causality-driven Adhoc Information Retrieval (CAIR-2021) shared task. The CAIR shared task expands traditional information retrieval to support the retrieval of documents containing the likely causes of a query event. A successful system must be able to distinguish between topical documents and documents containing causal descriptions of events that are causally related to the query event. Our approach involves aggregating results from multiple query strategies over a semantic and lexical index. The proposed approach leads the CAIR-2021 leaderboard and outperformed both traditional IR and pure semantic embedding-based approaches.
☆ Towards An Efficient LLM Training Paradigm for CTR Prediction
Large Language Models (LLMs) have demonstrated tremendous potential as the next-generation ranking-based recommendation system. Many recent works have shown that LLMs can significantly outperform conventional click-through-rate (CTR) prediction approaches. Despite such promising results, the computational inefficiency inherent in the current training paradigm makes it particularly challenging to train LLMs for ranking-based recommendation tasks on large datasets. To train LLMs for CTR prediction, most existing studies adopt the prevalent ''sliding-window'' paradigm. Given a sequence of $m$ user interactions, a unique training prompt is constructed for each interaction by designating it as the prediction target along with its preceding $n$ interactions serving as context. In turn, the sliding-window paradigm results in an overall complexity of $O(mn^2)$ that scales linearly with the length of user interactions. Consequently, a direct adoption to train LLMs with such strategy can result in prohibitively high training costs as the length of interactions grows. To alleviate the computational inefficiency, we propose a novel training paradigm, namely Dynamic Target Isolation (DTI), that structurally parallelizes the training of $k$ (where $k >> 1$) target interactions. Furthermore, we identify two major bottlenecks - hidden-state leakage and positional bias overfitting - that limit DTI to only scale up to a small value of $k$ (e.g., 5) then propose a computationally light solution to effectively tackle each. Through extensive experiments on three widely adopted public CTR datasets, we empirically show that DTI reduces training time by an average of $\textbf{92%}$ (e.g., from $70.5$ hrs to $5.31$ hrs), without compromising CTR prediction performance.
☆ Federated Conversational Recommender System ECIR 2024
Conversational Recommender Systems (CRSs) have become increasingly popular as a powerful tool for providing personalized recommendation experiences. By directly engaging with users in a conversational manner to learn their current and fine-grained preferences, a CRS can quickly derive recommendations that are relevant and justifiable. However, existing conversational recommendation systems (CRSs) typically rely on a centralized training and deployment process, which involves collecting and storing explicitly-communicated user preferences in a centralized repository. These fine-grained user preferences are completely human-interpretable and can easily be used to infer sensitive information (e.g., financial status, political stands, and health information) about the user, if leaked or breached. To address the user privacy concerns in CRS, we first define a set of privacy protection guidelines for preserving user privacy under the conversational recommendation setting. Based on these guidelines, we propose a novel federated conversational recommendation framework that effectively reduces the risk of exposing user privacy by (i) de-centralizing both the historical interests estimation stage and the interactive preference elicitation stage and (ii) strictly bounding privacy leakage by enforcing user-level differential privacy with meticulously selected privacy budgets. Through extensive experiments, we show that the proposed framework not only satisfies these user privacy protection guidelines, but also enables the system to achieve competitive recommendation performance even when compared to the state-of-the-art non-private conversational recommendation approach.
comment: ECIR 2024
☆ Systematic Literature Review on Clinical Trial Eligibility Matching
Clinical trial eligibility matching is a critical yet often labor-intensive and error-prone step in medical research, as it ensures that participants meet precise criteria for safe and reliable study outcomes. Recent advances in Natural Language Processing (NLP) have shown promise in automating and improving this process by rapidly analyzing large volumes of unstructured clinical text and structured electronic health record (EHR) data. In this paper, we present a systematic overview of current NLP methodologies applied to clinical trial eligibility screening, focusing on data sources, annotation practices, machine learning approaches, and real-world implementation challenges. A comprehensive literature search (spanning Google Scholar, Mendeley, and PubMed from 2015 to 2024) yielded high-quality studies, each demonstrating the potential of techniques such as rule-based systems, named entity recognition, contextual embeddings, and ontology-based normalization to enhance patient matching accuracy. While results indicate substantial improvements in screening efficiency and precision, limitations persist regarding data completeness, annotation consistency, and model scalability across diverse clinical domains. The review highlights how explainable AI and standardized ontologies can bolster clinician trust and broaden adoption. Looking ahead, further research into advanced semantic and temporal representations, expanded data integration, and rigorous prospective evaluations is necessary to fully realize the transformative potential of NLP in clinical trial recruitment.
☆ Towards Efficient Educational Chatbots: Benchmarking RAG Frameworks
Large Language Models (LLMs) have proven immensely beneficial in education by capturing vast amounts of literature-based information, allowing them to generate context without relying on external sources. In this paper, we propose a generative AI-powered GATE question-answering framework (GATE stands for Graduate Aptitude Test in Engineering) that leverages LLMs to explain GATE solutions and support students in their exam preparation. We conducted extensive benchmarking to select the optimal embedding model and LLM, evaluating our framework based on criteria such as latency, faithfulness, and relevance, with additional validation through human evaluation. Our chatbot integrates state-of-the-art embedding models and LLMs to deliver accurate, context-aware responses. Through rigorous experimentation, we identified configurations that balance performance and computational efficiency, ensuring a reliable chatbot to serve students' needs. Additionally, we discuss the challenges faced in data processing and modeling and implemented solutions. Our work explores the application of Retrieval-Augmented Generation (RAG) for GATE Q/A explanation tasks, and our findings demonstrate significant improvements in retrieval accuracy and response quality. This research offers practical insights for developing effective AI-driven educational tools while highlighting areas for future enhancement in usability and scalability.
☆ OrdRankBen: A Novel Ranking Benchmark for Ordinal Relevance in NLP
The evaluation of ranking tasks remains a significant challenge in natural language processing (NLP), particularly due to the lack of direct labels for results in real-world scenarios. Benchmark datasets play a crucial role in providing standardized testbeds that ensure fair comparisons, enhance reproducibility, and enable progress tracking, facilitating rigorous assessment and continuous improvement of ranking models. Existing NLP ranking benchmarks typically use binary relevance labels or continuous relevance scores, neglecting ordinal relevance scores. However, binary labels oversimplify relevance distinctions, while continuous scores lack a clear ordinal structure, making it challenging to capture nuanced ranking differences effectively. To address these challenges, we introduce OrdRankBen, a novel benchmark designed to capture multi-granularity relevance distinctions. Unlike conventional benchmarks, OrdRankBen incorporates structured ordinal labels, enabling more precise ranking evaluations. Given the absence of suitable datasets for ordinal relevance ranking in NLP, we constructed two datasets with distinct ordinal label distributions. We further evaluate various models for three model types, ranking-based language models, general large language models, and ranking-focused large language models on these datasets. Experimental results show that ordinal relevance modeling provides a more precise evaluation of ranking models, improving their ability to distinguish multi-granularity differences among ranked items-crucial for tasks that demand fine-grained relevance differentiation.
comment: 6 pages
☆ Optimizing Multi-Hop Document Retrieval Through Intermediate Representations
Retrieval-augmented generation (RAG) encounters challenges when addressing complex queries, particularly multi-hop questions. While several methods tackle multi-hop queries by iteratively generating internal queries and retrieving external documents, these approaches are computationally expensive. In this paper, we identify a three-stage information processing pattern in LLMs during layer-by-layer reasoning, consisting of extraction, processing, and subsequent extraction steps. This observation suggests that the representations in intermediate layers contain richer information compared to those in other layers. Building on this insight, we propose Layer-wise RAG (L-RAG). Unlike prior methods that focus on generating new internal queries, L-RAG leverages intermediate representations from the middle layers, which capture next-hop information, to retrieve external knowledge. L-RAG achieves performance comparable to multi-step approaches while maintaining inference overhead similar to that of standard RAG. Experimental results show that L-RAG outperforms existing RAG methods on open-domain multi-hop question-answering datasets, including MuSiQue, HotpotQA, and 2WikiMultiHopQA. The code is available in https://anonymous.4open.science/r/L-RAG-ADD5/
♻ ☆ VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents
Retrieval-augmented generation (RAG) is an effective technique that enables large language models (LLMs) to utilize external knowledge sources for generation. However, current RAG systems are solely based on text, rendering it impossible to utilize vision information like layout and images that play crucial roles in real-world multi-modality documents. In this paper, we introduce VisRAG, which tackles this issue by establishing a vision-language model (VLM)-based RAG pipeline. In this pipeline, instead of first parsing the document to obtain text, the document is directly embedded using a VLM as an image and then retrieved to enhance the generation of a VLM. Compared to traditional text-based RAG, VisRAG maximizes the retention and utilization of the data information in the original documents, eliminating the information loss introduced during the parsing process. We collect both open-source and synthetic data to train the retriever in VisRAG and explore a variety of generation methods. Experiments demonstrate that VisRAG outperforms traditional RAG in both the retrieval and generation stages, achieving a 20--40% end-to-end performance gain over traditional text-based RAG pipeline. Further analysis reveals that VisRAG is efficient in utilizing training data and demonstrates strong generalization capability, positioning it as a promising solution for RAG on multi-modality documents. Our code and data are available at https://github.com/openbmb/visrag.
Machine Learning 64
♻ ☆ Eagle: Exploring The Design Space for Multimodal LLMs with Mixture of Encoders
The ability to accurately interpret complex visual information is a crucial topic of multimodal large language models (MLLMs). Recent work indicates that enhanced visual perception significantly reduces hallucinations and improves performance on resolution-sensitive tasks, such as optical character recognition and document analysis. A number of recent MLLMs achieve this goal using a mixture of vision encoders. Despite their success, there is a lack of systematic comparisons and detailed ablation studies addressing critical aspects, such as expert selection and the integration of multiple vision experts. This study provides an extensive exploration of the design space for MLLMs using a mixture of vision encoders and resolutions. Our findings reveal several underlying principles common to various existing strategies, leading to a streamlined yet effective design approach. We discover that simply concatenating visual tokens from a set of complementary vision encoders is as effective as more complex mixing architectures or strategies. We additionally introduce Pre-Alignment to bridge the gap between vision-focused encoders and language tokens, enhancing model coherence. The resulting family of MLLMs, Eagle, surpasses other leading open-source models on major MLLM benchmarks.
comment: Github: https://github.com/NVlabs/Eagle, HuggingFace: https://huggingface.co/NVEagle
♻ ☆ SymbolFit: Automatic Parametric Modeling with Symbolic Regression
We introduce SymbolFit, a framework that automates parametric modeling by using symbolic regression to perform a machine-search for functions that fit the data while simultaneously providing uncertainty estimates in a single run. Traditionally, constructing a parametric model to accurately describe binned data has been a manual and iterative process, requiring an adequate functional form to be determined before the fit can be performed. The main challenge arises when the appropriate functional forms cannot be derived from first principles, especially when there is no underlying true closed-form function for the distribution. In this work, we develop a framework that automates and streamlines the process by utilizing symbolic regression, a machine learning technique that explores a vast space of candidate functions without requiring a predefined functional form because the functional form itself is treated as a trainable parameter, making the process far more efficient and effortless than traditional regression methods. We demonstrate the framework in high-energy physics experiments at the CERN Large Hadron Collider (LHC) using five real proton-proton collision datasets from new physics searches, including background modeling in resonance searches for high-mass dijet, trijet, paired-dijet, diphoton, and dimuon events. We show that our framework can flexibly and efficiently generate a wide range of candidate functions that fit a nontrivial distribution well using a simple fit configuration that varies only by random seed, and that the same fit configuration, which defines a vast function space, can also be applied to distributions of different shapes, whereas achieving a comparable result with traditional methods would have required extensive manual effort.
comment: 50 pages, 35 figures. Under review. The API can be used out-of-the-box and is available at https://github.com/hftsoi/symbolfit
♻ ☆ Quantum time dynamics mediated by the Yang-Baxter equation and artificial neural networks
Quantum computing shows great potential, but errors pose a significant challenge. This study explores new strategies for mitigating quantum errors using artificial neural networks (ANN) and the Yang-Baxter equation (YBE). Unlike traditional error mitigation methods, which are computationally intensive, we investigate artificial error mitigation. We developed a novel method that combines ANN for noise mitigation combined with the YBE to generate noisy data. This approach effectively reduces noise in quantum simulations, enhancing the accuracy of the results. The YBE rigorously preserves quantum correlations and symmetries in spin chain simulations in certain classes of integrable lattice models, enabling effective compression of quantum circuits while retaining linear scalability with the number of qubits. This compression facilitates both full and partial implementations, allowing the generation of noisy quantum data on hardware alongside noiseless simulations using classical platforms. By introducing controlled noise through the YBE, we enhance the dataset for error mitigation. We train an ANN model on partial data from quantum simulations, demonstrating its effectiveness in mitigating errors in time-evolving quantum states, providing a scalable framework to enhance quantum computation fidelity, particularly in noisy intermediate-scale quantum (NISQ) systems. We demonstrate the efficacy of this approach by performing quantum time dynamics simulations using the Heisenberg XY Hamiltonian on real quantum devices.
♻ ☆ Heterogeneous Graph Neural Network on Semantic Tree AAAI 2025
The recent past has seen an increasing interest in Heterogeneous Graph Neural Networks (HGNNs), since many real-world graphs are heterogeneous in nature, from citation graphs to email graphs. However, existing methods ignore a tree hierarchy among metapaths, naturally constituted by different node types and relation types. In this paper, we present HetTree, a novel HGNN that models both the graph structure and heterogeneous aspects in a scalable and effective manner. Specifically, HetTree builds a semantic tree data structure to capture the hierarchy among metapaths. To effectively encode the semantic tree, HetTree uses a novel subtree attention mechanism to emphasize metapaths that are more helpful in encoding parent-child relationships. Moreover, HetTree proposes carefully matching pre-computed features and labels correspondingly, constituting a complete metapath representation. Our evaluation of HetTree on a variety of real-world datasets demonstrates that it outperforms all existing baselines on open benchmarks and efficiently scales to large real-world graphs with millions of nodes and edges.
comment: Accepted at AAAI 2025
♻ ☆ Disentangling Representations through Multi-task Learning
Intelligent perception and interaction with the world hinges on internal representations that capture its underlying structure (''disentangled'' or ''abstract'' representations). Disentangled representations serve as world models, isolating latent factors of variation in the world along approximately orthogonal directions, thus facilitating feature-based generalization. We provide experimental and theoretical results guaranteeing the emergence of disentangled representations in agents that optimally solve multi-task evidence accumulation classification tasks, canonical in the neuroscience literature. The key conceptual finding is that, by producing accurate multi-task classification estimates, a system implicitly represents a set of coordinates specifying a disentangled representation of the underlying latent state of the data it receives. The theory provides conditions for the emergence of these representations in terms of noise, number of tasks, and evidence accumulation time. We experimentally validate these predictions in RNNs trained to multi-task, which learn disentangled representations in the form of continuous attractors, leading to zero-shot out-of-distribution (OOD) generalization in predicting latent factors. We demonstrate the robustness of our framework across autoregressive architectures, decision boundary geometries and in tasks requiring classification confidence estimation. We find that transformers are particularly suited for disentangling representations, which might explain their unique world understanding abilities. Overall, our framework establishes a formal link between competence at multiple tasks and the formation of disentangled, interpretable world models in both biological and artificial systems, and helps explain why ANNs often arrive at human-interpretable concepts, and how they both may acquire exceptional zero-shot generalization capabilities.
comment: 43 pages, 17 figures
♻ ☆ We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs USENIX Security
The reliance of popular programming languages such as Python and JavaScript on centralized package repositories and open-source software, combined with the emergence of code-generating Large Language Models (LLMs), has created a new type of threat to the software supply chain: package hallucinations. These hallucinations, which arise from fact-conflicting errors when generating code using LLMs, represent a novel form of package confusion attack that poses a critical threat to the integrity of the software supply chain. This paper conducts a rigorous and comprehensive evaluation of package hallucinations across different programming languages, settings, and parameters, exploring how a diverse set of models and configurations affect the likelihood of generating erroneous package recommendations and identifying the root causes of this phenomenon. Using 16 popular LLMs for code generation and two unique prompt datasets, we generate 576,000 code samples in two programming languages that we analyze for package hallucinations. Our findings reveal that that the average percentage of hallucinated packages is at least 5.2% for commercial models and 21.7% for open-source models, including a staggering 205,474 unique examples of hallucinated package names, further underscoring the severity and pervasiveness of this threat. To overcome this problem, we implement several hallucination mitigation strategies and show that they are able to significantly reduce the number of package hallucinations while maintaining code quality. Our experiments and findings highlight package hallucinations as a persistent and systemic phenomenon while using state-of-the-art LLMs for code generation, and a significant challenge which deserves the research community's urgent attention.
comment: To appear in the 2025 USENIX Security Symposium. 22 pages, 14 figures, 8 tables. Edited from original version for submission to a different conference. No change to original results or findings
♻ ☆ Range, not Independence, Drives Modularity in Biologically Inspired Representations
Why do biological and artificial neurons sometimes modularise, each encoding a single meaningful variable, and sometimes entangle their representation of many variables? In this work, we develop a theory of when biologically inspired networks -- those that are nonnegative and energy efficient -- modularise their representation of source variables (sources). We derive necessary and sufficient conditions on a sample of sources that determine whether the neurons in an optimal biologically-inspired linear autoencoder modularise. Our theory applies to any dataset, extending far beyond the case of statistical independence studied in previous work. Rather we show that sources modularise if their support is ``sufficiently spread''. From this theory, we extract and validate predictions in a variety of empirical studies on how data distribution affects modularisation in nonlinear feedforward and recurrent neural networks trained on supervised and unsupervised tasks. Furthermore, we apply these ideas to neuroscience data, showing that range independence can be used to understand the mixing or modularising of spatial and reward information in entorhinal recordings in seemingly conflicting experiments. Further, we use these results to suggest alternate origins of mixed-selectivity, beyond the predominant theory of flexible nonlinear classification. In sum, our theory prescribes precise conditions on when neural activities modularise, providing tools for inducing and elucidating modular representations in brains and machines.
comment: 47 pages, 17 figures. WD and KH contributed equally; LH and JHL contributed equally
♻ ☆ MetaGFN: Exploring Distant Modes with Adapted Metadynamics for Continuous GFlowNets
Generative Flow Networks (GFlowNets) are a class of generative models that sample objects in proportion to a specified reward function through a learned policy. They can be trained either on-policy or off-policy, needing a balance between exploration and exploitation for fast convergence to a target distribution. While exploration strategies for discrete GFlowNets have been studied, exploration in the continuous case remains to be investigated, despite the potential for novel exploration algorithms due to the local connectedness of continuous domains. Here, we introduce Adapted Metadynamics, a variant of metadynamics that can be applied to arbitrary black-box reward functions on continuous domains. We use Adapted Metadynamics as an exploration strategy for continuous GFlowNets. We show several continuous domains where the resulting algorithm, MetaGFN, accelerates convergence to the target distribution and discovers more distant reward modes than previous off-policy exploration strategies used for GFlowNets.
comment: 22 pages
♻ ☆ Linear Diffusion Networks
Diffusion kernels capture global dependencies. We present Linear Diffusion Networks (LDNs), a novel architecture that reinterprets sequential data processing as a unified diffusion process. Our model integrates adaptive diffusion modules with localized nonlinear updates and a diffusion-inspired attention mechanism. This design enables efficient global information propagation while preserving fine-grained temporal details. LDN overcomes the limitations of conventional recurrent and transformer models by allowing full parallelization across time steps and supporting robust multi-scale temporal representations. Experiments on benchmark sequence modeling tasks demonstrate that LDN delivers competitive performance across ImageNet and GLUE tasks.
♻ ☆ $μ$nit Scaling: Simple and Scalable FP8 LLM Training
Large Language Model training with 8-bit floating point (FP8) formats promises significant efficiency improvements, but reduced numerical precision makes training challenging. It is currently possible to train in FP8 only if one is willing to tune various hyperparameters, reduce model scale, or accept the overhead of computing dynamic scale factors. We demonstrate simple, scalable FP8 training that requires no dynamic scaling factors or special hyperparameters, even at large model sizes. Our method, $\mu$nit Scaling ($\mu$S), also enables simple hyperparameter transfer across model widths, matched numerics across training and inference, and other desirable properties. $\mu$nit Scaling is straightforward to implement, consisting of a set of minimal interventions based on a first-principles analysis of common transformer operations. We validate our method by training models from 1B to 13B parameters, performing all hidden linear layer computations in FP8. We achieve quality equal to higher precision baselines while also training up to 33% faster.
♻ ☆ Lean Copilot: Large Language Models as Copilots for Theorem Proving in Lean
Neural theorem proving combines large language models (LLMs) with proof assistants such as Lean, where the correctness of formal proofs can be rigorously verified, leaving no room for hallucination. With existing neural theorem provers pretrained on a fixed collection of data and offering valuable suggestions at times, it is challenging for them to continually prove novel theorems in a fully autonomous mode, where human insights may be critical. In this paper, we explore LLMs as copilots that assist humans in proving theorems. We introduce Lean Copilot, an general framework for running LLM inference natively in Lean. It enables programmers to build various LLM-based proof automation tools that integrate seamlessly into the workflow of Lean users. Lean users can use our pretrained models or bring their own ones that run either locally (with or without GPUs) or on the cloud. Using Lean Copilot, we build LLM-based tools that suggest proof steps, complete proof goals, and select relevant premises. Experimental results on the Mathematics in Lean textbook demonstrate the effectiveness of our method compared to existing rule-based proof automation in Lean (aesop). When assisting humans, Lean Copilot requires only 2.08 manually-entered proof steps on average (3.86 required by aesop); when automating the theorem proving process, Lean Copilot automates 74.2% proof steps on average, 85% better than aesop (40.1%). We open source all code and artifacts under a permissive MIT license to facilitate further research.
comment: All code and artifacts open-sourced at https://github.com/lean-dojo/LeanCopilot
♻ ☆ Fast Two-Time-Scale Stochastic Gradient Method with Applications in Reinforcement Learning
Two-time-scale optimization is a framework introduced in Zeng et al. (2024) that abstracts a range of policy evaluation and policy optimization problems in reinforcement learning (RL). Akin to bi-level optimization under a particular type of stochastic oracle, the two-time-scale optimization framework has an upper level objective whose gradient evaluation depends on the solution of a lower level problem, which is to find the root of a strongly monotone operator. In this work, we propose a new method for solving two-time-scale optimization that achieves significantly faster convergence than the prior arts. The key idea of our approach is to leverage an averaging step to improve the estimates of the operators in both lower and upper levels before using them to update the decision variables. These additional averaging steps eliminate the direct coupling between the main variables, enabling the accelerated performance of our algorithm. We characterize the finite-time convergence rates of the proposed algorithm under various conditions of the underlying objective function, including strong convexity, Polyak-Lojasiewicz condition, and general non-convexity. These rates significantly improve over the best-known complexity of the standard two-time-scale stochastic approximation algorithm. When applied to RL, we show how the proposed algorithm specializes to novel online sample-based methods that surpass or match the performance of the existing state of the art. Finally, we support our theoretical results with numerical simulations in RL.
♻ ☆ LDAdam: Adaptive Optimization from Low-Dimensional Gradient Statistics ICLR 2025
We introduce LDAdam, a memory-efficient optimizer for training large models, that performs adaptive optimization steps within lower dimensional subspaces, while consistently exploring the full parameter space during training. This strategy keeps the optimizer's memory footprint to a fraction of the model size. LDAdam relies on a new projection-aware update rule for the optimizer states that allows for transitioning between subspaces, i.e., estimation of the statistics of the projected gradients. To mitigate the errors due to low-rank projection, LDAdam integrates a new generalized error feedback mechanism, which explicitly accounts for both gradient and optimizer state compression. We prove the convergence of LDAdam under standard assumptions, and show that LDAdam allows for accurate and efficient fine-tuning and pre-training of language models. Code is available at https://github.com/IST-DASLab/LDAdam
comment: 39 pages, ICLR 2025
♻ ☆ Fréchet Wavelet Distance: A Domain-Agnostic Metric for Image Generation
Modern metrics for generative learning like Fr\'echet Inception Distance (FID) and DINOv2-Fr\'echet Distance (FD-DINOv2) demonstrate impressive performance. However, they suffer from various shortcomings, like a bias towards specific generators and datasets. To address this problem, we propose the Fr\'echet Wavelet Distance (FWD) as a domain-agnostic metric based on the Wavelet Packet Transform ($W_p$). FWD provides a sight across a broad spectrum of frequencies in images with a high resolution, preserving both spatial and textural aspects. Specifically, we use $W_p$ to project generated and real images to the packet coefficient space. We then compute the Fr\'echet distance with the resultant coefficients to evaluate the quality of a generator. This metric is general-purpose and dataset-domain agnostic, as it does not rely on any pre-trained network, while being more interpretable due to its ability to compute Fr\'echet distance per packet, enhancing transparency. We conclude with an extensive evaluation of a wide variety of generators across various datasets that the proposed FWD can generalize and improve robustness to domain shifts and various corruptions compared to other metrics.
♻ ☆ Kolmogorov-Arnold PointNet: Deep learning for prediction of fluid fields on irregular geometries
Kolmogorov-Arnold Networks (KANs) have emerged as a promising alternative to traditional Multilayer Perceptrons (MLPs) in deep learning. KANs have already been integrated into various architectures, such as convolutional neural networks, graph neural networks, and transformers, and their potential has been assessed for predicting physical quantities. However, the combination of KANs with point-cloud-based neural networks (e.g., PointNet) for computational physics has not yet been explored. To address this, we present Kolmogorov-Arnold PointNet (KA-PointNet) as a novel supervised deep learning framework for the prediction of incompressible steady-state fluid flow fields in irregular domains, where the predicted fields are a function of the geometry of the domains. In KA-PointNet, we implement shared KANs in the segmentation branch of the PointNet architecture. We utilize Jacobi polynomials to construct shared KANs. As a benchmark test case, we consider incompressible laminar steady-state flow over a cylinder, where the geometry of its cross-section varies over the data set. We investigate the performance of Jacobi polynomials with different degrees as well as special cases of Jacobi polynomials such as Legendre polynomials, Chebyshev polynomials of the first and second kinds, and Gegenbauer polynomials, in terms of the computational cost of training and accuracy of prediction of the test set. Additionally, we compare the performance of PointNet with shared KANs (i.e., KA-PointNet) and PointNet with shared MLPs. It is observed that when the number of trainable parameters is approximately equal, PointNet with shared KANs (i.e., KA-PointNet) outperforms PointNet with shared MLPs. Moreover, KA-PointNet predicts the pressure and velocity distributions along the surface of cylinders more accurately, resulting in more precise computations of lift and drag.
♻ ☆ Distributed Speculative Inference (DSI): Speculation Parallelism for Provably Faster Lossless Language Model Inference ICLR 2025
This paper introduces distributed speculative inference (DSI), a novel inference algorithm that is provably faster than speculative inference (SI) [leviathan2023, chen2023, miao2024, sun2025, timor2025] and standard autoregressive inference (non-SI). Like other SI algorithms, DSI operates on frozen language models (LMs), requiring no training or architectural modifications, and it preserves the target distribution. Prior studies on SI have demonstrated empirical speedups over non-SI--but rely on sufficiently fast and accurate drafters, which are often unavailable in practice. We identify a gap where SI can be slower than non-SI if drafters are too slow or inaccurate. We close this gap by proving that DSI is faster than both SI and non-SI--given any drafters. DSI is therefore not only faster than SI, but also unlocks the acceleration of LMs for which SI fails. DSI leverages speculation parallelism (SP), a novel type of task parallelism, to orchestrate target and drafter instances that overlap in time, establishing a new foundational tradeoff between computational resources and latency. Our simulations show that DSI is 1.29-1.92x faster than SI in single-node setups for various off-the-shelf LMs and tasks. We open-source all our code.
comment: Published at ICLR 2025. (Link: https://openreview.net/forum?id=cJd1BgZ9CS)
♻ ☆ TESGNN: Temporal Equivariant Scene Graph Neural Networks for Efficient and Robust Multi-View 3D Scene Understanding
Scene graphs have proven to be highly effective for various scene understanding tasks due to their compact and explicit representation of relational information. However, current methods often overlook the critical importance of preserving symmetry when generating scene graphs from 3D point clouds, which can lead to reduced accuracy and robustness, particularly when dealing with noisy, multi-view data. Furthermore, a major limitation of prior approaches is the lack of temporal modeling to capture time-dependent relationships among dynamically evolving entities in a scene. To address these challenges, we propose Temporal Equivariant Scene Graph Neural Network (TESGNN), consisting of two key components: (1) an Equivariant Scene Graph Neural Network (ESGNN), which extracts information from 3D point clouds to generate scene graph while preserving crucial symmetry properties, and (2) a Temporal Graph Matching Network, which fuses scene graphs generated by ESGNN across multiple time sequences into a unified global representation using an approximate graph-matching algorithm. Our combined architecture TESGNN outperforms current state-of-the-art methods in scene graph generation, achieving higher accuracy and faster training convergence. Moreover, we show that leveraging the symmetry-preserving property produces a more stable and accurate global scene representation compared to existing approaches. Last but not least, it is computationally efficient and easily implementable using existing frameworks, making it well-suited for real-time applications in robotics and computer vision. This approach paves the way for more robust and scalable solutions to complex multi-view scene understanding challenges. Our source code is publicly available at: https://github.com/HySonLab/TESGraph
comment: arXiv admin note: text overlap with arXiv:2407.00609
♻ ☆ Unmasking Social Bots: How Confident Are We?
Social bots remain a major vector for spreading disinformation on social media and a menace to the public. Despite the progress made in developing multiple sophisticated social bot detection algorithms and tools, bot detection remains a challenging, unsolved problem that is fraught with uncertainty due to the heterogeneity of bot behaviors, training data, and detection algorithms. Detection models often disagree on whether to label the same account as bot or human-controlled. However, they do not provide any measure of uncertainty to indicate how much we should trust their results. We propose to address both bot detection and the quantification of uncertainty at the account level - a novel feature of this research. This dual focus is crucial as it allows us to leverage additional information related to the quantified uncertainty of each prediction, thereby enhancing decision-making and improving the reliability of bot classifications. Specifically, our approach facilitates targeted interventions for bots when predictions are made with high confidence and suggests caution (e.g., gathering more data) when predictions are uncertain.
comment: 15 pages, 6 figures, 6 tables
♻ ☆ The Optimization Landscape of SGD Across the Feature Learning Strength ICLR 2025
We consider neural networks (NNs) where the final layer is down-scaled by a fixed hyperparameter $\gamma$. Recent work has identified $\gamma$ as controlling the strength of feature learning. As $\gamma$ increases, network evolution changes from "lazy" kernel dynamics to "rich" feature-learning dynamics, with a host of associated benefits including improved performance on common tasks. In this work, we conduct a thorough empirical investigation of the effect of scaling $\gamma$ across a variety of models and datasets in the online training setting. We first examine the interaction of $\gamma$ with the learning rate $\eta$, identifying several scaling regimes in the $\gamma$-$\eta$ plane which we explain theoretically using a simple model. We find that the optimal learning rate $\eta^*$ scales non-trivially with $\gamma$. In particular, $\eta^* \propto \gamma^2$ when $\gamma \ll 1$ and $\eta^* \propto \gamma^{2/L}$ when $\gamma \gg 1$ for a feed-forward network of depth $L$. Using this optimal learning rate scaling, we proceed with an empirical study of the under-explored "ultra-rich" $\gamma \gg 1$ regime. We find that networks in this regime display characteristic loss curves, starting with a long plateau followed by a drop-off, sometimes followed by one or more additional staircase steps. We find networks of different large $\gamma$ values optimize along similar trajectories up to a reparameterization of time. We further find that optimal online performance is often found at large $\gamma$ and could be missed if this hyperparameter is not tuned. Our findings indicate that analytical study of the large-$\gamma$ limit may yield useful insights into the dynamics of representation learning in performant models.
comment: ICLR 2025 Final Copy, 40 Pages, 45 figures
♻ ☆ Detecting Unsuccessful Students in Cybersecurity Exercises in Two Different Learning Environments
This full paper in the research track evaluates the usage of data logged from cybersecurity exercises in order to predict students who are potentially at risk of performing poorly. Hands-on exercises are essential for learning since they enable students to practice their skills. In cybersecurity, hands-on exercises are often complex and require knowledge of many topics. Therefore, students may miss solutions due to gaps in their knowledge and become frustrated, which impedes their learning. Targeted aid by the instructor helps, but since the instructor's time is limited, efficient ways to detect struggling students are needed. This paper develops automated tools to predict when a student is having difficulty. We formed a dataset with the actions of 313 students from two countries and two learning environments: KYPO CRP and EDURange. These data are used in machine learning algorithms to predict the success of students in exercises deployed in these environments. After extracting features from the data, we trained and cross-validated eight classifiers for predicting the exercise outcome and evaluated their predictive power. The contribution of this paper is comparing two approaches to feature engineering, modeling, and classification performance on data from two learning environments. Using the features from either learning environment, we were able to detect and distinguish between successful and struggling students. A decision tree classifier achieved the highest balanced accuracy and sensitivity with data from both learning environments. The results show that activity data from cybersecurity exercises are suitable for predicting student success. In a potential application, such models can aid instructors in detecting struggling students and providing targeted help. We publish data and code for building these models so that others can adopt or adapt them.
comment: Published in the FIE 2024 conference proceedings, see https://doi.org/10.1109/FIE61694.2024.10893135
♻ ☆ Differentiable Weightless Neural Networks
We introduce the Differentiable Weightless Neural Network (DWN), a model based on interconnected lookup tables. Training of DWNs is enabled by a novel Extended Finite Difference technique for approximate differentiation of binary values. We propose Learnable Mapping, Learnable Reduction, and Spectral Regularization to further improve the accuracy and efficiency of these models. We evaluate DWNs in three edge computing contexts: (1) an FPGA-based hardware accelerator, where they demonstrate superior latency, throughput, energy efficiency, and model area compared to state-of-the-art solutions, (2) a low-power microcontroller, where they achieve preferable accuracy to XGBoost while subject to stringent memory constraints, and (3) ultra-low-cost chips, where they consistently outperform small models in both accuracy and projected hardware area. DWNs also compare favorably against leading approaches for tabular datasets, with higher average rank. Overall, our work positions DWNs as a pioneering solution for edge-compatible high-throughput neural networks.
♻ ☆ Learning General-Purpose Biomedical Volume Representations using Randomized Synthesis ICLR 2025
Current volumetric biomedical foundation models struggle to generalize as public 3D datasets are small and do not cover the broad diversity of medical procedures, conditions, anatomical regions, and imaging protocols. We address this by creating a representation learning method that instead anticipates strong domain shifts at training time itself. We first propose a data engine that synthesizes highly variable training samples that would enable generalization to new biomedical contexts. To then train a single 3D network for any voxel-level task, we develop a contrastive learning method that pretrains the network to be stable against nuisance imaging variation simulated by the data engine, a key inductive bias for generalization. This network's features can be used as robust representations of input images for downstream tasks and its weights provide a strong, dataset-agnostic initialization for finetuning on new datasets. As a result, we set new standards across both multimodality registration and few-shot segmentation, a first for any 3D biomedical vision model, all without (pre-)training on any existing dataset of real images.
comment: ICLR 2025: International Conference on Learning Representations. Code and model weights available at https://github.com/neel-dey/anatomix. Keywords: synthetic data, representation learning, medical image analysis, image registration, image segmentation
♻ ☆ Prompting Fairness: Integrating Causality to Debias Large Language Models
Large language models (LLMs), despite their remarkable capabilities, are susceptible to generating biased and discriminatory responses. As LLMs increasingly influence high-stakes decision-making (e.g., hiring and healthcare), mitigating these biases becomes critical. In this work, we propose a causality-guided debiasing framework to tackle social biases, aiming to reduce the objectionable dependence between LLMs' decisions and the social information in the input. Our framework introduces a novel perspective to identify how social information can affect an LLM's decision through different causal pathways. Leveraging these causal insights, we outline principled prompting strategies that regulate these pathways through selection mechanisms. This framework not only unifies existing prompting-based debiasing techniques, but also opens up new directions for reducing bias by encouraging the model to prioritize fact-based reasoning over reliance on biased social cues. We validate our framework through extensive experiments on real-world datasets across multiple domains, demonstrating its effectiveness in debiasing LLM decisions, even with only black-box access to the model.
comment: 24 pages, 10 figures
♻ ☆ Explain Yourself, Briefly! Self-Explaining Neural Networks with Concise Sufficient Reasons ICLR 2025
*Minimal sufficient reasons* represent a prevalent form of explanation - the smallest subset of input features which, when held constant at their corresponding values, ensure that the prediction remains unchanged. Previous *post-hoc* methods attempt to obtain such explanations but face two main limitations: (1) Obtaining these subsets poses a computational challenge, leading most scalable methods to converge towards suboptimal, less meaningful subsets; (2) These methods heavily rely on sampling out-of-distribution input assignments, potentially resulting in counterintuitive behaviors. To tackle these limitations, we propose in this work a self-supervised training approach, which we term *sufficient subset training* (SST). Using SST, we train models to generate concise sufficient reasons for their predictions as an integral part of their output. Our results indicate that our framework produces succinct and faithful subsets substantially more efficiently than competing post-hoc methods, while maintaining comparable predictive performance.
comment: To appear in ICLR 2025
♻ ☆ SymDiff: Equivariant Diffusion via Stochastic Symmetrisation ICLR 2025
We propose SymDiff, a method for constructing equivariant diffusion models using the framework of stochastic symmetrisation. SymDiff resembles a learned data augmentation that is deployed at sampling time, and is lightweight, computationally efficient, and easy to implement on top of arbitrary off-the-shelf models. In contrast to previous work, SymDiff typically does not require any neural network components that are intrinsically equivariant, avoiding the need for complex parameterisations or the use of higher-order geometric features. Instead, our method can leverage highly scalable modern architectures as drop-in replacements for these more constrained alternatives. We show that this additional flexibility yields significant empirical benefit for $\mathrm{E}(3)$-equivariant molecular generation. To the best of our knowledge, this is the first application of symmetrisation to generative modelling, suggesting its potential in this domain more generally.
comment: Camera-ready version for ICLR 2025
♻ ☆ FedBiP: Heterogeneous One-Shot Federated Learning with Personalized Latent Diffusion Models CVPR 2025
One-Shot Federated Learning (OSFL), a special decentralized machine learning paradigm, has recently gained significant attention. OSFL requires only a single round of client data or model upload, which reduces communication costs and mitigates privacy threats compared to traditional FL. Despite these promising prospects, existing methods face challenges due to client data heterogeneity and limited data quantity when applied to real-world OSFL systems. Recently, Latent Diffusion Models (LDM) have shown remarkable advancements in synthesizing high-quality images through pretraining on large-scale datasets, thereby presenting a potential solution to overcome these issues. However, directly applying pretrained LDM to heterogeneous OSFL results in significant distribution shifts in synthetic data, leading to performance degradation in classification models trained on such data. This issue is particularly pronounced in rare domains, such as medical imaging, which are underrepresented in LDM's pretraining data. To address this challenge, we propose Federated Bi-Level Personalization (FedBiP), which personalizes the pretrained LDM at both instance-level and concept-level. Hereby, FedBiP synthesizes images following the client's local data distribution without compromising the privacy regulations. FedBiP is also the first approach to simultaneously address feature space heterogeneity and client data scarcity in OSFL. Our method is validated through extensive experiments on three OSFL benchmarks with feature space heterogeneity, as well as on challenging medical and satellite image datasets with label heterogeneity. The results demonstrate the effectiveness of FedBiP, which substantially outperforms other OSFL methods.
comment: CVPR 2025
♻ ☆ Towards Understanding the Universality of Transformers for Next-Token Prediction ICLR 2025
Causal Transformers are trained to predict the next token for a given context. While it is widely accepted that self-attention is crucial for encoding the causal structure of sequences, the precise underlying mechanism behind this in-context autoregressive learning ability remains unclear. In this paper, we take a step towards understanding this phenomenon by studying the approximation ability of Transformers for next-token prediction. Specifically, we explore the capacity of causal Transformers to predict the next token $x_{t+1}$ given an autoregressive sequence $(x_1, \dots, x_t)$ as a prompt, where $ x_{t+1} = f(x_t) $, and $ f $ is a context-dependent function that varies with each sequence. On the theoretical side, we focus on specific instances, namely when $ f $ is linear or when $ (x_t)_{t \geq 1} $ is periodic. We explicitly construct a Transformer (with linear, exponential, or softmax attention) that learns the mapping $f$ in-context through a causal kernel descent method. The causal kernel descent method we propose provably estimates $x_{t+1} $ based solely on past and current observations $ (x_1, \dots, x_t) $, with connections to the Kaczmarz algorithm in Hilbert spaces. We present experimental results that validate our theoretical findings and suggest their applicability to more general mappings $f$.
comment: ICLR 2025, 20 pages
♻ ☆ Leveraging Dual Process Theory in Language Agent Framework for Real-time Simultaneous Human-AI Collaboration
Agents built on large language models (LLMs) have excelled in turn-by-turn human-AI collaboration but struggle with simultaneous tasks requiring real-time interaction. Latency issues and the challenge of inferring variable human strategies hinder their ability to make autonomous decisions without explicit instructions. Through experiments with current independent System 1 and System 2 methods, we validate the necessity of using Dual Process Theory (DPT) in real-time tasks. We propose DPT-Agent, a novel language agent framework that integrates System 1 and System 2 for efficient real-time simultaneous human-AI collaboration. DPT-Agent's System 1 uses a Finite-state Machine (FSM) and code-as-policy for fast, intuitive, and controllable decision-making. DPT-Agent's System 2 integrates Theory of Mind (ToM) and asynchronous reflection to infer human intentions and perform reasoning-based autonomous decisions. We demonstrate the effectiveness of DPT-Agent through further experiments with rule-based agents and human collaborators, showing significant improvements over mainstream LLM-based frameworks. DPT-Agent can effectively help LLMs convert correct slow thinking and reasoning into executable actions, thereby improving performance. To the best of our knowledge, DPT-Agent is the first language agent framework that achieves successful real-time simultaneous human-AI collaboration autonomously. Code of DPT-Agent can be found in https://github.com/sjtu-marl/DPT-Agent.
comment: Preprint under review. Update the experimental results of the DeepSeek-R1 series models, o3-mini-high and o3-mini-medium
♻ ☆ Bidirectional Consistency Models ICML 2024
Diffusion models (DMs) are capable of generating remarkably high-quality samples by iteratively denoising a random vector, a process that corresponds to moving along the probability flow ordinary differential equation (PF ODE). Interestingly, DMs can also invert an input image to noise by moving backward along the PF ODE, a key operation for downstream tasks such as interpolation and image editing. However, the iterative nature of this process restricts its speed, hindering its broader application. Recently, Consistency Models (CMs) have emerged to address this challenge by approximating the integral of the PF ODE, largely reducing the number of iterations. Yet, the absence of an explicit ODE solver complicates the inversion process. To resolve this, we introduce Bidirectional Consistency Model (BCM), which learns a single neural network that enables both forward and backward traversal along the PF ODE, efficiently unifying generation and inversion tasks within one framework. We can train BCM from scratch or tune it using a pretrained consistency model, which reduces the training cost and increases scalability. We demonstrate that BCM enables one-step generation and inversion while also allowing the use of additional steps to enhance generation quality or reduce reconstruction error. We further showcase BCM's capability in downstream tasks, such as interpolation and inpainting. Our code and weights are available at https://github.com/Mosasaur5526/BCM-iCT-torch.
comment: 39 pages, 27 figures; a shorter version of this paper was acceppted at the ICML 2024 Workshop on Structured Probabilistic Inference & Generative Modeling
♻ ☆ MoE-CAP: Benchmarking Cost, Accuracy and Performance of Sparse Mixture-of-Experts Systems
The Mixture-of-Experts (MoE) architecture is increasingly favored for scaling Large Language Models (LLMs). Its key feature, sparse activation, selectively activates only a subset of parameters (experts) per token, reducing memory bandwidth and compute FLOPs compared to dense models. To capitalize on this, MoE designers leverage heterogeneous compute and memory hardware to lower system costs. However, the interaction between model sparsity and hardware heterogeneity introduces trade-offs in Cost, Accuracy, and Performance (CAP). To address this, we introduce MoE-CAP, a benchmarking method for evaluating sparse MoE systems across these three dimensions. Its key innovation is a sparsity-aware CAP analysis model, the first to integrate cost, performance, and accuracy metrics into a single diagram while estimating the impact of sparsity on system performance. MoE-CAP helps practitioners optimize hardware provisioning for an MoE model-or vice versa. MoE-CAP supports various MoE models and provides more accurate metrics than existing methods.
♻ ☆ Foundational Policy Acquisition via Multitask Learning for Motor Skill Generation
In this study, we propose a multitask reinforcement learning algorithm for foundational policy acquisition to generate novel motor skills. \textcolor{\hcolor}{Learning the rich representation of the multitask policy is a challenge in dynamic movement generation tasks because the policy needs to cope with changes in goals or environments with different reward functions or physical parameters. Inspired by human sensorimotor adaptation mechanisms, we developed the learning pipeline to construct the encoder-decoder networks and network selection to facilitate foundational policy acquisition under multiple situations. First, we compared the proposed method with previous multitask reinforcement learning methods in the standard multi-locomotion tasks. The results showed that the proposed approach outperformed the baseline methods. Then, we applied the proposed method to the ball heading task using a monopod robot model to evaluate skill generation performance. The results showed that the proposed method was able to adapt to novel target positions or inexperienced ball restitution coefficients but to acquire a foundational policy network, originally learned for heading motion, which can generate an entirely new overhead kicking skill.
comment: 11 pages, 9 figures
♻ ☆ Phase-Amplitude Reduction-Based Imitation Learning
In this study, we propose the use of the phase-amplitude reduction method to construct an imitation learning framework. Imitating human movement trajectories is recognized as a promising strategy for generating a range of human-like robot movements. Unlike previous dynamical system-based imitation learning approaches, our proposed method allows the robot not only to imitate a limit cycle trajectory but also to replicate the transient movement from the initial or disturbed state to the limit cycle. Consequently, our method offers a safer imitation learning approach that avoids generating unpredictable motions immediately after disturbances or from a specified initial state. We first validated our proposed method by reconstructing a simple limit-cycle attractor. We then compared the proposed approach with a conventional method on a lemniscate trajectory tracking task with a simulated robot arm. Our findings confirm that our proposed method can more accurately generate transient movements to converge on a target periodic attractor compared to the previous standard approach. Subsequently, we applied our method to a real robot arm to imitate periodic human movements.
comment: 21 pages, 8 figures
♻ ☆ Nonparametric Heterogeneous Long-term Causal Effect Estimation via Data Combination
Long-term causal inference has drawn increasing attention in many scientific domains. Existing methods mainly focus on estimating average long-term causal effects by combining long-term observational data and short-term experimental data. However, it is still understudied how to robustly and effectively estimate heterogeneous long-term causal effects, significantly limiting practical applications. In this paper, we propose several two-stage style nonparametric estimators for heterogeneous long-term causal effect estimation, including propensity-based, regression-based, and multiple robust estimators. We conduct a comprehensive theoretical analysis of their asymptotic properties under mild assumptions, with the ultimate goal of building a better understanding of the conditions under which some estimators can be expected to perform better. Extensive experiments across several semi-synthetic and real-world datasets validate the theoretical results and demonstrate the effectiveness of the proposed estimators.
♻ ☆ TradingAgents: Multi-Agents LLM Financial Trading Framework AAAI 2025
Significant progress has been made in automated problem-solving using societies of agents powered by large language models (LLMs). In finance, efforts have largely focused on single-agent systems handling specific tasks or multi-agent frameworks independently gathering data. However, multi-agent systems' potential to replicate real-world trading firms' collaborative dynamics remains underexplored. TradingAgents proposes a novel stock trading framework inspired by trading firms, featuring LLM-powered agents in specialized roles such as fundamental analysts, sentiment analysts, technical analysts, and traders with varied risk profiles. The framework includes Bull and Bear researcher agents assessing market conditions, a risk management team monitoring exposure, and traders synthesizing insights from debates and historical data to make informed decisions. By simulating a dynamic, collaborative trading environment, this framework aims to improve trading performance. Detailed architecture and extensive experiments reveal its superiority over baseline models, with notable improvements in cumulative returns, Sharpe ratio, and maximum drawdown, highlighting the potential of multi-agent LLM frameworks in financial trading. TradingAgents is available at https://github.com/PioneerFintech.
comment: Multi-Agent AI in the Real World @ AAAI 2025
♻ ☆ Video-Foley: Two-Stage Video-To-Sound Generation via Temporal Event Condition For Foley Sound
Foley sound synthesis is crucial for multimedia production, enhancing user experience by synchronizing audio and video both temporally and semantically. Recent studies on automating this labor-intensive process through video-to-sound generation face significant challenges. Systems lacking explicit temporal features suffer from poor alignment and controllability, while timestamp-based models require costly and subjective human annotation. We propose Video-Foley, a video-to-sound system using Root Mean Square (RMS) as an intuitive condition with semantic timbre prompts (audio or text). RMS, a frame-level intensity envelope closely related to audio semantics, acts as a temporal event feature to guide audio generation from video. The annotation-free self-supervised learning framework consists of two stages, Video2RMS and RMS2Sound, incorporating novel ideas including RMS discretization and RMS-ControlNet with a pretrained text-to-audio model. Our extensive evaluation shows that Video-Foley achieves state-of-the-art performance in audio-visual alignment and controllability for sound timing, intensity, timbre, and nuance. Source code, model weights and demos are available on our companion website. (https://jnwnlee.github.io/video-foley-demo)
♻ ☆ Audio-Visual Instance Segmentation CVPR 2025
In this paper, we propose a new multi-modal task, termed audio-visual instance segmentation (AVIS), which aims to simultaneously identify, segment and track individual sounding object instances in audible videos. To facilitate this research, we introduce a high-quality benchmark named AVISeg, containing over 90K instance masks from 26 semantic categories in 926 long videos. Additionally, we propose a strong baseline model for this task. Our model first localizes sound source within each frame, and condenses object-specific contexts into concise tokens. Then it builds long-range audio-visual dependencies between these tokens using window-based attention, and tracks sounding objects among the entire video sequences. Extensive experiments reveal that our method performs best on AVISeg, surpassing the existing methods from related tasks. We further conduct the evaluation on several multi-modal large models. Unfortunately, they exhibits subpar performance on instance-level sound source localization and temporal perception. We expect that AVIS will inspire the community towards a more comprehensive multi-modal understanding. Dataset and code is available at https://github.com/ruohaoguo/avis.
comment: Accepted by CVPR 2025
♻ ☆ GAMED: Knowledge Adaptive Multi-Experts Decoupling for Multimodal Fake News Detection
Multimodal fake news detection often involves modelling heterogeneous data sources, such as vision and language. Existing detection methods typically rely on fusion effectiveness and cross-modal consistency to model the content, complicating understanding how each modality affects prediction accuracy. Additionally, these methods are primarily based on static feature modelling, making it difficult to adapt to the dynamic changes and relationships between different data modalities. This paper develops a significantly novel approach, GAMED, for multimodal modelling, which focuses on generating distinctive and discriminative features through modal decoupling to enhance cross-modal synergies, thereby optimizing overall performance in the detection process. GAMED leverages multiple parallel expert networks to refine features and pre-embed semantic knowledge to improve the experts' ability in information selection and viewpoint sharing. Subsequently, the feature distribution of each modality is adaptively adjusted based on the respective experts' opinions. GAMED also introduces a novel classification technique to dynamically manage contributions from different modalities, while improving the explainability of decisions. Experimental results on the Fakeddit and Yang datasets demonstrate that GAMED performs better than recently developed state-of-the-art models. The source code can be accessed at https://github.com/slz0925/GAMED.
♻ ☆ Improving LSH via Tensorized Random Projection
Locality sensitive hashing (LSH) is a fundamental algorithmic toolkit used by data scientists for approximate nearest neighbour search problems that have been used extensively in many large scale data processing applications such as near duplicate detection, nearest neighbour search, clustering, etc. In this work, we aim to propose faster and space efficient locality sensitive hash functions for Euclidean distance and cosine similarity for tensor data. Typically, the naive approach for obtaining LSH for tensor data involves first reshaping the tensor into vectors, followed by applying existing LSH methods for vector data $E2LSH$ and $SRP$. However, this approach becomes impractical for higher order tensors because the size of the reshaped vector becomes exponential in the order of the tensor. Consequently, the size of LSH parameters increases exponentially. To address this problem, we suggest two methods for LSH for Euclidean distance and cosine similarity, namely $CP-E2LSH$, $TT-E2LSH$, and $CP-SRP$, $TT-SRP$, respectively, building on $CP$ and tensor train $(TT)$ decompositions techniques. Our approaches are space efficient and can be efficiently applied to low rank $CP$ or $TT$ tensors. We provide a rigorous theoretical analysis of our proposal on their correctness and efficacy.
♻ ☆ Where is the Testbed for my Federated Learning Research?
Progressing beyond centralized AI is of paramount importance, yet, distributed AI solutions, in particular various federated learning (FL) algorithms, are often not comprehensively assessed, which prevents the research community from identifying the most promising approaches and practitioners from being convinced that a certain solution is deployment-ready. The largest hurdle towards FL algorithm evaluation is the difficulty of conducting real-world experiments over a variety of FL client devices and different platforms, with different datasets and data distribution, all while assessing various dimensions of algorithm performance, such as inference accuracy, energy consumption, and time to convergence, to name a few. In this paper, we present CoLExT, a real-world testbed for FL research. CoLExT is designed to streamline experimentation with custom FL algorithms in a rich testbed configuration space, with a large number of heterogeneous edge devices, ranging from single-board computers to smartphones, and provides real-time collection and visualization of a variety of metrics through automatic instrumentation. According to our evaluation, porting FL algorithms to CoLExT requires minimal involvement from the developer, and the instrumentation introduces minimal resource usage overhead. Furthermore, through an initial investigation involving popular FL algorithms running on CoLExT, we reveal previously unknown trade-offs, inefficiencies, and programming bugs.
comment: SEC 2024
♻ ☆ When Attention Sink Emerges in Language Models: An Empirical View ICLR 2025
Language Models (LMs) assign significant attention to the first token, even if it is not semantically important, which is known as attention sink. This phenomenon has been widely adopted in applications such as streaming/long context generation, KV cache optimization, inference acceleration, model quantization, and others. Despite its widespread use, a deep understanding of attention sink in LMs is still lacking. In this work, we first demonstrate that attention sinks exist universally in LMs with various inputs, even in small models. Furthermore, attention sink is observed to emerge during the LM pre-training, motivating us to investigate how optimization, data distribution, loss function, and model architecture in LM pre-training influence its emergence. We highlight that attention sink emerges after effective optimization on sufficient training data. The sink position is highly correlated with the loss function and data distribution. Most importantly, we find that attention sink acts more like key biases, storing extra attention scores, which could be non-informative and not contribute to the value computation. We also observe that this phenomenon (at least partially) stems from tokens' inner dependence on attention scores as a result of softmax normalization. After relaxing such dependence by replacing softmax attention with other attention operations, such as sigmoid attention without normalization, attention sinks do not emerge in LMs up to 1B parameters. The code is available at https://github.com/sail-sg/Attention-Sink.
comment: ICLR 2025 (Spotlight)
♻ ☆ Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates ICLR 2025
Automatic LLM benchmarks, such as AlpacaEval 2.0, Arena-Hard-Auto, and MT-Bench, have become popular for evaluating language models due to their cost-effectiveness and scalability compared to human evaluation. Achieving high win rates on these benchmarks can significantly boost the promotional impact of newly released language models. This promotional benefit may motivate tricks, such as manipulating model output length or style to game win rates, even though several mechanisms have been developed to control length and disentangle style to reduce gameability. Nonetheless, we show that even a "null model" that always outputs a constant response (irrelevant to input instructions) can cheat automatic benchmarks and achieve top-ranked win rates: an 86.5% LC win rate on AlpacaEval 2.0; an 83.0 score on Arena-Hard-Auto; and a 9.55 score on MT-Bench. Moreover, the crafted cheating outputs are transferable because we assume that the instructions of these benchmarks (e.g., 805 samples of AlpacaEval 2.0) are private and cannot be accessed. While our experiments are primarily proof-of-concept, an adversary could use LLMs to generate more imperceptible cheating responses, unethically benefiting from high win rates and promotional impact. Our findings call for the development of anti-cheating mechanisms for reliable automatic benchmarks. The code is available at https://github.com/sail-sg/Cheating-LLM-Benchmarks.
comment: ICLR 2025 (Oral)
♻ ☆ Graph Transformers Dream of Electric Flow
We show theoretically and empirically that the linear Transformer, when applied to graph data, can implement algorithms that solve canonical problems such as electric flow and eigenvector decomposition. The Transformer has access to information on the input graph only via the graph's incidence matrix. We present explicit weight configurations for implementing each algorithm, and we bound the constructed Transformers' errors by the errors of the underlying algorithms. Our theoretical findings are corroborated by experiments on synthetic data. Additionally, on a real-world molecular regression task, we observe that the linear Transformer is capable of learning a more effective positional encoding than the default one based on Laplacian eigenvectors. Our work is an initial step towards elucidating the inner-workings of the Transformer for graph data. Code is available at https://github.com/chengxiang/LinearGraphTransformer
♻ ☆ Accelerating 3D Molecule Generation via Jointly Geometric Optimal Transport ICLR 2025
This paper proposes a new 3D molecule generation framework, called GOAT, for fast and effective 3D molecule generation based on the flow-matching optimal transport objective. Specifically, we formulate a geometric transport formula for measuring the cost of mapping multi-modal features (e.g., continuous atom coordinates and categorical atom types) between a base distribution and a target data distribution. Our formula is solved within a joint, equivariant, and smooth representation space. This is achieved by transforming the multi-modal features into a continuous latent space with equivariant networks. In addition, we find that identifying optimal distributional coupling is necessary for fast and effective transport between any two distributions. We further propose a mechanism for estimating and purifying optimal coupling to train the flow model with optimal transport. By doing so, GOAT can turn arbitrary distribution couplings into new deterministic couplings, leading to an estimated optimal transport plan for fast 3D molecule generation. The purification filters out the subpar molecules to ensure the ultimate generation quality. We theoretically and empirically prove that the proposed optimal coupling estimation and purification yield transport plan with non-increasing cost. Finally, extensive experiments show that GOAT enjoys the efficiency of solving geometric optimal transport, leading to a double speedup compared to the sub-optimal method while achieving the best generation quality regarding validity, uniqueness, and novelty. The code is available at https://github.com/WanyuGroup/ICLR2025-GOAT.
comment: Published as a conference paper at ICLR 2025
♻ ☆ Knowledge Gradient for Multi-Objective Bayesian Optimization with Decoupled Evaluations
Multi-objective Bayesian optimization aims to find the Pareto front of trade-offs between a set of expensive objectives while collecting as few samples as possible. In some cases, it is possible to evaluate the objectives separately, and a different latency or evaluation cost can be associated with each objective. This decoupling of the objectives presents an opportunity to learn the Pareto front faster by avoiding unnecessary, expensive evaluations. We propose a scalarization based knowledge gradient acquisition function which accounts for the different evaluation costs of the objectives. We prove asymptotic consistency of the estimator of the optimum for an arbitrary, D-dimensional, real compact search space and show empirically that the algorithm performs comparably with the state of the art and significantly outperforms versions which always evaluate both objectives.
comment: 36 pages. This preprint has not undergone peer review (when applicable) or any post-submission improvements or corrections. The Version of Record of this contribution is published in 'Evolutionary Multi-Criterion Optimization', LNCS 15513, and is available online at https://doi.org/10.1007/978-981-96-3538-2_9
♻ ☆ UMGAD: Unsupervised Multiplex Graph Anomaly Detection
Graph anomaly detection (GAD) is a critical task in graph machine learning, with the primary objective of identifying anomalous nodes that deviate significantly from the majority. This task is widely applied in various real-world scenarios, including fraud detection and social network analysis. However, existing GAD methods still face two major challenges: (1) They are often limited to detecting anomalies in single-type interaction graphs and struggle with multiple interaction types in multiplex heterogeneous graphs. (2) In unsupervised scenarios, selecting appropriate anomaly score thresholds remains a significant challenge for accurate anomaly detection. To address the above challenges, we propose a novel Unsupervised Multiplex Graph Anomaly Detection method, named UMGAD. We first learn multi-relational correlations among nodes in multiplex heterogeneous graphs and capture anomaly information during node attribute and structure reconstruction through graph-masked autoencoder (GMAE). Then, to further extract abnormal information, we generate attribute-level and subgraph-level augmented-view graphs respectively, and perform attribute and structure reconstruction through GMAE. Finally, we learn to optimize node attributes and structural features through contrastive learning between original-view and augmented-view graphs to improve the model's ability to capture anomalies. Meanwhile, we also propose a new anomaly score threshold selection strategy, which allows the model to be independent of ground truth information in real unsupervised scenarios. Extensive experiments on four datasets show that our UMGAD significantly outperforms state-of-the-art methods, achieving average improvements of 13.48% in AUC and 11.68% in Macro-F1 across all datasets.
♻ ☆ MOVE: Effective and Harmless Ownership Verification via Embedded External Features AAAI 2022
Currently, deep neural networks (DNNs) are widely adopted in different applications. Despite its commercial values, training a well-performing DNN is resource-consuming. Accordingly, the well-trained model is valuable intellectual property for its owner. However, recent studies revealed the threats of model stealing, where the adversaries can obtain a function-similar copy of the victim model, even when they can only query the model. In this paper, we propose an effective and harmless model ownership verification (MOVE) to defend against different types of model stealing simultaneously, without introducing new security risks. In general, we conduct the ownership verification by verifying whether a suspicious model contains the knowledge of defender-specified external features. Specifically, we embed the external features by modifying a few training samples with style transfer. We then train a meta-classifier to determine whether a model is stolen from the victim. This approach is inspired by the understanding that the stolen models should contain the knowledge of features learned by the victim model. In particular, \revision{we develop our MOVE method under both white-box and black-box settings and analyze its theoretical foundation to provide comprehensive model protection.} Extensive experiments on benchmark datasets verify the effectiveness of our method and its resistance to potential adaptive attacks. The codes for reproducing the main experiments of our method are available at https://github.com/THUYimingLi/MOVE.
comment: This paper has been accepted by IEEE TPAMI 2025. It is the journal extension of our conference paper in AAAI 2022 (https://ojs.aaai.org/index.php/AAAI/article/view/20036). 18 pages
♻ ☆ Permutation-Invariant Graph Partitioning:How Graph Neural Networks Capture Structural Interactions?
Graph Neural Networks (GNNs) have paved the way for being a cornerstone in graph-related learning tasks. Yet, the ability of GNNs to capture structural interactions within graphs remains under-explored. In this work, we address this gap by drawing on the insight that permutation invariant graph partitioning enables a powerful way of exploring structural interactions. We establish theoretical connections between permutation invariant graph partitioning and graph isomorphism, and then propose Graph Partitioning Neural Networks (GPNNs), a novel architecture that efficiently enhances the expressive power of GNNs in learning structural interactions. We analyze how partitioning schemes and structural interactions contribute to GNN expressivity and their trade-offs with complexity. Empirically, we demonstrate that GPNNs outperform existing GNN models in capturing structural interactions across diverse graph benchmark tasks.
♻ ☆ Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations ICLR 2025
To gain insight into the mechanisms behind machine learning methods, it is crucial to establish connections among the features describing data points. However, these correlations often exhibit a high-dimensional and strongly nonlinear nature, which makes them challenging to detect using standard methods. This paper exploits the entanglement between intrinsic dimensionality and correlation to propose a metric that quantifies the (potentially nonlinear) correlation between high-dimensional manifolds. We first validate our method on synthetic data in controlled environments, showcasing its advantages and drawbacks compared to existing techniques. Subsequently, we extend our analysis to large-scale applications in neural network representations. Specifically, we focus on latent representations of multimodal data, uncovering clear correlations between paired visual and textual embeddings, whereas existing methods struggle significantly in detecting similarity. Our results indicate the presence of highly nonlinear correlation patterns between latent manifolds.
comment: Accepted at ICLR 2025
♻ ☆ Boosting Jailbreak Attack with Momentum ICASSP 2025
Large Language Models (LLMs) have achieved remarkable success across diverse tasks, yet they remain vulnerable to adversarial attacks, notably the well-known jailbreak attack. In particular, the Greedy Coordinate Gradient (GCG) attack has demonstrated efficacy in exploiting this vulnerability by optimizing adversarial prompts through a combination of gradient heuristics and greedy search. However, the efficiency of this attack has become a bottleneck in the attacking process. To mitigate this limitation, in this paper we rethink the generation of the adversarial prompts through an optimization lens, aiming to stabilize the optimization process and harness more heuristic insights from previous optimization iterations. Specifically, we propose the \textbf{M}omentum \textbf{A}ccelerated G\textbf{C}G (\textbf{MAC}) attack, which integrates a momentum term into the gradient heuristic to boost and stabilize the random search for tokens in adversarial prompts. Experimental results showcase the notable enhancement achieved by MAC over baselines in terms of attack success rate and optimization efficiency. Moreover, we demonstrate that MAC can still exhibit superior performance for transfer attacks and models under defense mechanisms. Our code is available at https://github.com/weizeming/momentum-attack-llm.
comment: Accepted by ICASSP 2025
♻ ☆ PIG: Physics-Informed Gaussians as Adaptive Parametric Mesh Representations
The numerical approximation of partial differential equations (PDEs) using neural networks has seen significant advancements through Physics-Informed Neural Networks (PINNs). Despite their straightforward optimization framework and flexibility in implementing various PDEs, PINNs often suffer from limited accuracy due to the spectral bias of Multi-Layer Perceptrons (MLPs), which struggle to effectively learn high-frequency and nonlinear components. Recently, parametric mesh representations in combination with neural networks have been investigated as a promising approach to eliminate the inductive bias of MLPs. However, they usually require high-resolution grids and a large number of collocation points to achieve high accuracy while avoiding overfitting. In addition, the fixed positions of the mesh parameters restrict their flexibility, making accurate approximation of complex PDEs challenging. To overcome these limitations, we propose Physics-Informed Gaussians (PIGs), which combine feature embeddings using Gaussian functions with a lightweight neural network. Our approach uses trainable parameters for the mean and variance of each Gaussian, allowing for dynamic adjustment of their positions and shapes during training. This adaptability enables our model to optimally approximate PDE solutions, unlike models with fixed parameter positions. Furthermore, the proposed approach maintains the same optimization framework used in PINNs, allowing us to benefit from their excellent properties. Experimental results show the competitive performance of our model across various PDEs, demonstrating its potential as a robust tool for solving complex PDEs. Our project page is available at https://namgyukang.github.io/Physics-Informed-Gaussians/
comment: Project page: https://namgyukang.github.io/Physics-Informed-Gaussians/
♻ ☆ Geometric Inductive Biases of Deep Networks: The Role of Data and Architecture
In this paper, we propose the $\textit{geometric invariance hypothesis (GIH)}$, which argues that the input space curvature of a neural network remains invariant under transformation in certain architecture-dependent directions during training. We investigate a simple, non-linear binary classification problem residing on a plane in a high dimensional space and observe that$\unicode{x2014}$unlike MPLs$\unicode{x2014}$ResNets fail to generalize depending on the orientation of the plane. Motivated by this example, we define a neural network's $\textbf{average geometry}$ and $\textbf{average geometry evolution}$ as compact $\textit{architecture-dependent}$ summaries of the model's input-output geometry and its evolution during training. By investigating the average geometry evolution at initialization, we discover that the geometry of a neural network evolves according to the data covariance projected onto its average geometry. This means that the geometry only changes in a subset of the input space when the average geometry is low-rank, such as in ResNets. This causes an architecture-dependent invariance property in the input space curvature, which we dub GIH. Finally, we present extensive experimental results to observe the consequences of GIH and how it relates to generalization in neural networks.
♻ ☆ On Theoretical Limits of Learning with Label Differential Privacy
Label differential privacy (DP) is designed for learning problems involving private labels and public features. While various methods have been proposed for learning under label DP, the theoretical limits remain largely unexplored. In this paper, we investigate the fundamental limits of learning with label DP in both local and central models for both classification and regression tasks, characterized by minimax convergence rates. We establish lower bounds by converting each task into a multiple hypothesis testing problem and bounding the test error. Additionally, we develop algorithms that yield matching upper bounds. Our results demonstrate that under label local DP (LDP), the risk has a significantly faster convergence rate than that under full LDP, i.e. protecting both features and labels, indicating the advantages of relaxing the DP definition to focus solely on labels. In contrast, under the label central DP (CDP), the risk is only reduced by a constant factor compared to full DP, indicating that the relaxation of CDP only has limited benefits on the performance.
♻ ☆ FOSP: Fine-tuning Offline Safe Policy through World Models ICLR2025
Offline Safe Reinforcement Learning (RL) seeks to address safety constraints by learning from static datasets and restricting exploration. However, these approaches heavily rely on the dataset and struggle to generalize to unseen scenarios safely. In this paper, we aim to improve safety during the deployment of vision-based robotic tasks through online fine-tuning an offline pretrained policy. To facilitate effective fine-tuning, we introduce model-based RL, which is known for its data efficiency. Specifically, our method employs in-sample optimization to improve offline training efficiency while incorporating reachability guidance to ensure safety. After obtaining an offline safe policy, a safe policy expansion approach is leveraged for online fine-tuning. The performance of our method is validated on simulation benchmarks with five vision-only tasks and through real-world robot deployment using limited data. It demonstrates that our approach significantly improves the generalization of offline policies to unseen safety-constrained scenarios. To the best of our knowledge, this is the first work to explore offline-to-online RL for safe generalization tasks.
comment: 32 pages, ICLR2025
♻ ☆ Breaking the Reclustering Barrier in Centroid-based Deep Clustering ICLR 2025
This work investigates an important phenomenon in centroid-based deep clustering (DC) algorithms: Performance quickly saturates after a period of rapid early gains. Practitioners commonly address early saturation with periodic reclustering, which we demonstrate to be insufficient to address performance plateaus. We call this phenomenon the "reclustering barrier" and empirically show when the reclustering barrier occurs, what its underlying mechanisms are, and how it is possible to Break the Reclustering Barrier with our algorithm BRB. BRB avoids early over-commitment to initial clusterings and enables continuous adaptation to reinitialized clustering targets while remaining conceptually simple. Applying our algorithm to widely-used centroid-based DC algorithms, we show that (1) BRB consistently improves performance across a wide range of clustering benchmarks, (2) BRB enables training from scratch, and (3) BRB performs competitively against state-of-the-art DC algorithms when combined with a contrastive loss. We release our code and pre-trained models at https://github.com/Probabilistic-and-Interactive-ML/breaking-the-reclustering-barrier .
comment: Accepted at ICLR 2025 (Camera-ready version)
♻ ☆ Greener GRASS: Enhancing GNNs with Encoding, Rewiring, and Attention ICLR 2025
Graph Neural Networks (GNNs) have become important tools for machine learning on graph-structured data. In this paper, we explore the synergistic combination of graph encoding, graph rewiring, and graph attention, by introducing Graph Attention with Stochastic Structures (GRASS), a novel GNN architecture. GRASS utilizes relative random walk probabilities (RRWP) encoding and a novel decomposed variant (D-RRWP) to efficiently capture structural information. It rewires the input graph by superimposing a random regular graph to enhance long-range information propagation. It also employs a novel additive attention mechanism tailored for graph-structured data. Our empirical evaluations demonstrate that GRASS achieves state-of-the-art performance on multiple benchmark datasets, including a 20.3% reduction in mean absolute error on the ZINC dataset.
comment: Published as a conference paper at ICLR 2025
♻ ☆ Robust Weight Initialization for Tanh Neural Networks with Fixed Point Analysis ICLR 2025
As a neural network's depth increases, it can improve generalization performance. However, training deep networks is challenging due to gradient and signal propagation issues. To address these challenges, extensive theoretical research and various methods have been introduced. Despite these advances, effective weight initialization methods for tanh neural networks remain insufficiently investigated. This paper presents a novel weight initialization method for neural networks with tanh activation function. Based on an analysis of the fixed points of the function $\tanh(ax)$, the proposed method aims to determine values of $a$ that mitigate activation saturation. A series of experiments on various classification datasets and physics-informed neural networks demonstrates that the proposed method outperforms Xavier initialization methods~(with or without normalization) in terms of robustness across different network sizes, data efficiency, and convergence speed. Code is available at https://github.com/1HyunwooLee/Tanh-Init
comment: ICLR 2025
♻ ☆ Taxonomy, Opportunities, and Challenges of Representation Engineering for Large Language Models
Representation Engineering (RepE) is a novel paradigm for controlling the behavior of LLMs. Unlike traditional approaches that modify inputs or fine-tune the model, RepE directly manipulates the model's internal representations. As a result, it may offer more effective, interpretable, data-efficient, and flexible control over models' behavior. We present the first comprehensive survey of RepE for LLMs, reviewing the rapidly growing literature to address key questions: What RepE methods exist and how do they differ? For what concepts and problems has RepE been applied? What are the strengths and weaknesses of RepE compared to other methods? To answer these, we propose a unified framework describing RepE as a pipeline comprising representation identification, operationalization, and control. We posit that while RepE methods offer significant potential, challenges remain, including managing multiple concepts, ensuring reliability, and preserving models' performance. Towards improving RepE, we identify opportunities for experimental and methodological improvements and construct a guide for best practices.
♻ ☆ Timer-XL: Long-Context Transformers for Unified Time Series Forecasting
We present Timer-XL, a causal Transformer for unified time series forecasting. To uniformly predict multidimensional time series, we generalize next token prediction, predominantly adopted for 1D token sequences, to multivariate next token prediction. The paradigm formulates various forecasting tasks as a long-context prediction problem. We opt for decoder-only Transformers that capture causal dependencies from varying-length contexts for unified forecasting, making predictions on non-stationary univariate time series, multivariate series with complicated dynamics and correlations, as well as covariate-informed contexts that include exogenous variables. Technically, we propose a universal TimeAttention to capture fine-grained intra- and inter-series dependencies of flattened time series tokens (patches), which is further enhanced by deft position embedding for temporal causality and variable equivalence. Timer-XL achieves state-of-the-art performance across task-specific forecasting benchmarks through a unified approach. Based on large-scale pre-training, Timer-XL achieves state-of-the-art zero-shot performance, making it a promising architecture for pre-trained time series models. Code is available at this repository: https://github.com/thuml/Timer-XL.
♻ ☆ On the Asymptotic Mean Square Error Optimality of Diffusion Models
Diffusion models (DMs) as generative priors have recently shown great potential for denoising tasks but lack theoretical understanding with respect to their mean square error (MSE) optimality. This paper proposes a novel denoising strategy inspired by the structure of the MSE-optimal conditional mean estimator (CME). The resulting DM-based denoiser can be conveniently employed using a pre-trained DM, being particularly fast by truncating reverse diffusion steps and not requiring stochastic re-sampling. We present a comprehensive (non-)asymptotic optimality analysis of the proposed diffusion-based denoiser, demonstrating polynomial-time convergence to the CME under mild conditions. Our analysis also derives a novel Lipschitz constant that depends solely on the DM's hyperparameters. Further, we offer a new perspective on DMs, showing that they inherently combine an asymptotically optimal denoiser with a powerful generator, modifiable by switching re-sampling in the reverse process on or off. The theoretical findings are thoroughly validated with experiments based on various benchmark datasets
♻ ☆ End-to-End Modeling Hierarchical Time Series Using Autoregressive Transformer and Conditional Normalizing Flow based Reconciliation ICDM2022
Multivariate time series forecasting with hierarchical structure is pervasive in real-world applications, demanding not only predicting each level of the hierarchy, but also reconciling all forecasts to ensure coherency, i.e., the forecasts should satisfy the hierarchical aggregation constraints. Moreover, the disparities of statistical characteristics between levels can be huge, worsened by non-Gaussian distributions and non-linear correlations. To this extent, we propose a novel end-to-end hierarchical time series forecasting model, based on conditioned normalizing flow-based autoregressive transformer reconciliation, to represent complex data distribution while simultaneously reconciling the forecasts to ensure coherency. Unlike other state-of-the-art methods, we achieve the forecasting and reconciliation simultaneously without requiring any explicit post-processing step. In addition, by harnessing the power of deep model, we do not rely on any assumption such as unbiased estimates or Gaussian distribution. Our evaluation experiments are conducted on four real-world hierarchical datasets from different industrial domains (three public ones and a dataset from the application servers of Alipay's data center) and the preliminary results demonstrate efficacy of our proposed method.
comment: Accepted by the 22nd IEEE International Conference on Data Mining (ICDM2022)
♻ ☆ Training-Free Message Passing for Learning on Hypergraphs
Hypergraphs are crucial for modelling higher-order interactions in real-world data. Hypergraph neural networks (HNNs) effectively utilise these structures by message passing to generate informative node features for various downstream tasks like node classification. However, the message passing module in existing HNNs typically requires a computationally intensive training process, which limits their practical use. To tackle this challenge, we propose an alternative approach by decoupling the usage of hypergraph structural information from the model learning stage. This leads to a novel training-free message passing module, named TF-MP-Module, which can be precomputed in the data preprocessing stage, thereby reducing the computational burden. We refer to the hypergraph neural network equipped with our TF-MP-Module as TF-HNN. We theoretically support the efficiency and effectiveness of TF-HNN by showing that: 1) It is more training-efficient compared to existing HNNs; 2) It utilises as much information as existing HNNs for node feature generation; and 3) It is robust against the oversmoothing issue while using long-range interactions. Experiments based on seven real-world hypergraph benchmarks in node classification and hyperlink prediction show that, compared to state-of-the-art HNNs, TF-HNN exhibits both competitive performance and superior training efficiency. Specifically, on the large-scale benchmark, Trivago, TF-HNN outperforms the node classification accuracy of the best baseline by 10% with just 1% of the training time of that baseline.
♻ ☆ DiscoGraMS: Enhancing Movie Screen-Play Summarization using Movie Character-Aware Discourse Graph NAACL 2025
Summarizing movie screenplays presents a unique set of challenges compared to standard document summarization. Screenplays are not only lengthy, but also feature a complex interplay of characters, dialogues, and scenes, with numerous direct and subtle relationships and contextual nuances that are difficult for machine learning models to accurately capture and comprehend. Recent attempts at screenplay summarization focus on fine-tuning transformer-based pre-trained models, but these models often fall short in capturing long-term dependencies and latent relationships, and frequently encounter the "lost in the middle" issue. To address these challenges, we introduce DiscoGraMS, a novel resource that represents movie scripts as a movie character-aware discourse graph (CaD Graph). This approach is well-suited for various downstream tasks, such as summarization, question-answering, and salience detection. The model aims to preserve all salient information, offering a more comprehensive and faithful representation of the screenplay's content. We further explore a baseline method that combines the CaD Graph with the corresponding movie script through a late fusion of graph and text modalities, and we present very initial promising results.
comment: Accepted at NAACL 2025 (Main)
♻ ☆ Dist Loss: Enhancing Regression in Few-Shot Region through Distribution Distance Constraint
Imbalanced data distributions are prevalent in real-world scenarios, posing significant challenges in both imbalanced classification and imbalanced regression tasks. They often cause deep learning models to overfit in areas of high sample density (many-shot regions) while underperforming in areas of low sample density (few-shot regions). This characteristic restricts the utility of deep learning models in various sectors, notably healthcare, where areas with few-shot data hold greater clinical relevance. While recent studies have shown the benefits of incorporating distribution information in imbalanced classification tasks, such strategies are rarely explored in imbalanced regression. In this paper, we address this issue by introducing a novel loss function, termed Dist Loss, designed to minimize the distribution distance between the model's predictions and the target labels in a differentiable manner, effectively integrating distribution information into model training. Dist Loss enables deep learning models to regularize their output distribution during training, effectively enhancing their focus on few-shot regions. We have conducted extensive experiments across three datasets spanning computer vision and healthcare: IMDB-WIKI-DIR, AgeDB-DIR, and ECG-Ka-DIR. The results demonstrate that Dist Loss effectively mitigates the negative impact of imbalanced data distribution on model performance, achieving state-of-the-art results in sparse data regions. Furthermore, Dist Loss is easy to integrate, complementing existing methods.
♻ ☆ Market-Derived Financial Sentiment Analysis: Context-Aware Language Models for Crypto Forecasting
Financial Sentiment Analysis (FSA) traditionally relies on human-annotated sentiment labels to infer investor sentiment and forecast market movements. However, inferring the potential market impact of words based on their human-perceived intentions is inherently challenging. We hypothesize that the historical market reactions to words, offer a more reliable indicator of their potential impact on markets than subjective sentiment interpretations by human annotators. To test this hypothesis, a market-derived labeling approach is proposed to assign tweet labels based on ensuing short-term price trends, enabling the language model to capture the relationship between textual signals and market dynamics directly. A domain-specific language model was fine-tuned on these labels, achieving up to an 11% improvement in short-term trend prediction accuracy over traditional sentiment-based benchmarks. Moreover, by incorporating market and temporal context through prompt-tuning, the proposed context-aware language model demonstrated an accuracy of 89.6% on a curated dataset of 227 impactful Bitcoin-related news events with significant market impacts. Aggregating daily tweet predictions into trading signals, our method outperformed traditional fusion models (which combine sentiment-based and price-based predictions). It challenged the assumption that sentiment-based signals are inferior to price-based predictions in forecasting market movements. Backtesting these signals across three distinct market regimes yielded robust Sharpe ratios of up to 5.07 in trending markets and 3.73 in neutral markets. Our findings demonstrate that language models can serve as effective short-term market predictors. This paradigm shift underscores the untapped capabilities of language models in financial decision-making and opens new avenues for market prediction applications.
comment: 13 pages, 6 figures